OFFICIAL
GenCost 2025-26
Final report
Paul Graham, Jenny Hayward and James Foster
July 2026
Australia’s National
Science Agency
GenCost 2025-26 | i
Contact
Paul Graham
+61 2 4960 6061
paul.graham@csiro.au
Citation
Graham, P., Hayward, J. and Foster, J. 2026, GenCost 2025-26: Final report, CSIRO, Australia.
Acknowledgement
CSIRO acknowledges the Traditional Owners of the lands that we live and work on across Australia and pays its respect to Elders past and present
Copyright
© Commonwealth Scientific and Industrial Research Organisation 2026. To the extent permitted by law, all rights are reserved and no part of this publication covered by copyright may be reproduced or copied in any form or by any means except with the written permission of CSIRO.
Important disclaimer
CSIRO advises that the information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it.
CSIRO is committed to providing web accessible content wherever possible. If you are having difficulties with accessing this document please contact www.csiro.au/en/contact.
ii | CSIRO Australia’s National Science Agency
Contents
Foreword vii
Acknowledgements ....................................................................................................................... viii
Executive summary ......................................................................................................................... ix
1 Introduction ...................................................................................................................... 14
1.1 Scope of the GenCost project and reporting....................................................... 14
1.2 The GenCost mailing list ...................................................................................... 15
1.3 Summary of feedback received and changes to the consultation draft ............. 15
2 Current technology costs .................................................................................................. 19
2.1 Current cost definition ........................................................................................ 19
2.2 Capital cost source............................................................................................... 21
2.3 Current generation technology capital costs ...................................................... 21
2.4 Current storage technology capital costs ............................................................ 23
3 Scenario narratives and data assumptions ....................................................................... 26
3.1 Scenario narratives .............................................................................................. 26
4 Projection results .............................................................................................................. 28
4.1 Short-term and long-term inflationary pressures ............................................... 28
4.2 Global generation mix ......................................................................................... 30
4.3 Changes in capital cost projections ..................................................................... 32
5 Levelised cost of electricity analysis ................................................................................. 51
5.1 LCOE definition .................................................................................................... 51
5.2 Change in method for estimating the cost of reliable high VRE share generation 51
5.3 The 2030 SLCOE, existing capacity and electricity futures prices ....................... 52
5.4 2050 SLCOE scenarios .......................................................................................... 53
5.5 SLCOE estimates .................................................................................................. 55
5.6 Overall implications for long term electricity generation cost trends ................ 65 Global and local learning model .......................................................................... 66
Data tables ........................................................................................................... 69
Data assumptions ................................................................................................ 82
Frequently asked questions................................................................................. 89
GenCost 2025-26 | iii
Technology inclusion principles......................................................................... 102
Shortened forms ......................................................................................................................... 105
References ........................................................................................................................... 108
iv | CSIRO Australia’s National Science Agency
Figures
Figure 2-1 Comparison of current capital cost estimates with previous reports (in real terms) . 22
Figure 2-2 Year on year change in current capital costs of selected technologies in the past four years (in real terms) ...................................................................................................................... 22
Figure 2-3 Capital costs of storage technologies in $/kWh (total cost basis) ............................... 24
Figure 2-4 Capital costs of storage technologies in $/kW (total cost basis) ................................. 25
Figure 4-1 Projected global electricity generation mix in 2030 and 2050 by scenario ................ 31
Figure 4-2 Global hydrogen production by technology and scenario, Mt .................................... 32
Figure 4-3 Projected capital costs for black coal ultra-supercritical by scenario compared to 2024-25 projections ...................................................................................................................... 33
Figure 4-4 Projected capital costs for black coal with CCS by scenario compared to 2024-25 projections .................................................................................................................................... 34
Figure 4-5 Projected capital costs for gas combined cycle by scenario compared to 2024-25 projections .................................................................................................................................... 35
Figure 4-6 Projected capital costs for gas with CCS by scenario compared to 2024-25 projections ....................................................................................................................................................... 36
Figure 4-7 Projected capital costs for gas open cycle (small) by scenario compared to 2024-25 projections .................................................................................................................................... 37
Figure 4-8 Projected capital costs for nuclear SMR by scenario compared to 2024-25 projections ....................................................................................................................................................... 38
Figure 4-9 Projected capital costs for large-scale nuclear by scenario compared to 2024-25 projections .................................................................................................................................... 39
Figure 4-10 Projected capital costs for solar thermal with 14 hours storage compared to 2024-25 projections ............................................................................................................................... 40
Figure 4-11 Projected capital costs for large-scale solar PV by scenario compared to 2024-25 projections .................................................................................................................................... 41
Figure 4-12 Projected capital costs for rooftop solar PV by scenario compared to 2024-25 projections .................................................................................................................................... 42
Figure 4-13 Projected capital costs for onshore wind by scenario compared to 2024-25 projections .................................................................................................................................... 43
Figure 4-14 Projected capital costs for fixed and floating offshore wind by scenario compared to 2024-25 projections ...................................................................................................................... 44
Figure 4-15 Projected total capital costs for 2-hour duration batteries by scenario (battery and balance of plant) ........................................................................................................................... 45
Figure 4-16 Projected capital costs for pumped hydro energy storage (24-hour) by scenario ... 46
GenCost 2025-26 | v
Figure 4-17 Projected technology capital costs under the Current policies scenario compared to 2024-25 projections ...................................................................................................................... 47
Figure 4-18 Projected technology capital costs under the Global NZE by 2050 scenario compared to 2024-25 projections ................................................................................................ 48
Figure 4-19 Projected technology capital costs under the Global NZE post 2050 scenario compared to 2024-25 projections ................................................................................................ 49
Figure 4-20 Projected technology capital costs for alkaline and PEM electrolysers by scenario, compared to 2024-25 ................................................................................................................... 50
Figure 5-1 Baseload and Peak evening electricity futures prices by state, nominal $/MWh ....... 56
Figure 5-2 Calculated 2030 LCOE of peaking and bulk energy technologies compared to Baseload and Peak evening electricity futures price ranges ........................................................ 57
Figure 5-3 Historical and two year projection of capital cost of 4 hour duration battery and large open cycle gas turbine .................................................................................................................. 58
Figure 5-4 The projected generation mix in 2050 by emissions intensity target and allowed technology. .................................................................................................................................... 59
Figure 5-5 The projected SLCOE in 2050 in the NEM by emission intensity target and technology allowed .......................................................................................................................................... 61
Figure 5-6 The breakdown of SLCOE in 2050 in the NEM by cost component for mature technology only scenario .............................................................................................................. 62
Figure 5-7 Average and marginal cost of abatement to achieve lower emissions intensity targets in 2050 compared to whole of economy abatement costs .......................................................... 63
Figure 5-8 Calculated LCOE range by technology and SLCOE range for 2050 .............................. 65
Apx Figure A.1 Schematic of changes in the learning rate as a technology progresses through its development stages after commercialisation .............................................................................. 67
Tables
Table 2-1 Suggested FOAK premium by technology ..................................................................... 21
Table 3-1 Summary of scenarios and their key assumptions ....................................................... 27
Table 5-1 2050 full fuel cycle greenhouse gas emissions intensity target scenarios and their meaning......................................................................................................................................... 54
Apx Table A.1 Cost breakdown of offshore wind ......................................................................... 68
Apx Table B.1 Current and projected generation technology capital costs under the Current policies scenario ............................................................................................................................ 70
vi | CSIRO Australia’s National Science Agency
Apx Table B.2 Current and projected generation technology capital costs under the Global NZE by 2050 scenario ........................................................................................................................... 71
Apx Table B.3 Current and projected generation technology capital costs under the Global NZE post 2050 scenario ........................................................................................................................ 72
Apx Table B.4 One- and two-hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW) .......................................................................... 73
Apx Table B.5 Four- and eight-hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW) .......................................................................... 74
Apx Table B.6 Twelve- and twenty-four hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW) ........................................................... 75
Apx Table B.7 Pumped hydro storage cost data by duration, by scenario, total cost basis ......... 76
Apx Table B.8 Historical storage cost data, total cost basis ......................................................... 77
Apx Table B.9 Data assumptions for LCOE calculations ................................................................ 78
Apx Table B.10 Electricity generation technology LCOE projections data, $/MWh ..................... 80
Apx Table B.11 Hydrogen electrolyser cost projections by scenario and technology, $/kW ....... 81
Apx Table C.1 Assumed technology learning rates that vary by scenario .................................... 82
Apx Table C.2 Assumed technology learning rates that are the same under all scenarios .......... 84
Apx Table C.3 Hydrogen demand assumptions by scenario in 2050 ............................................ 86
Apx Table C.4 Maximum renewable generation shares in the year 2050, except for offshore wind which is in GW of installed capacity. .................................................................................... 87
Apx Table E.1 Examples of considering global or domestic significance .................................... 103
GenCost 2025-26 | vii
Foreword
Assumptions about the cost of electricity generation and storage technologies are a key input to any electricity system planning exercise in Australia or around the world. The primary role of GenCost is to provide capital cost data for the electricity modelling and planning community. The project delivers the capital cost data with an emphasis on stakeholder consultation, recognising that no single organisation can be completely across the changing circumstances of all relevant technologies.
A secondary goal of the project is to provide an indicator of what the capital cost data means for the cost of delivered electricity and the relative competitiveness of generation technologies. This function is delivered by calculating a metric called the levelised cost of electricity (LCOE) which is the minimum per unit price that a project requires to pay back its investment and running costs over its life. LCOE typically only consider a small number of core project details with the more minor or unique costs of each project ignored so that costs are calculated on a simple and common basis.
GenCost also provides a system levelised cost of electricity (SLCOE) which is the average cost of electricity from a bundle of electricity generation and storage technologies that together meet all the requirements of the electricity system. Electricity systems will always require a diversity of resources to deliver all their functions and so no single technology will meet all the system’s needs regardless of its relative cost position.
viii | CSIRO Australia’s National Science Agency
Acknowledgements
This report has benefited from feedback provided by electricity sector stakeholders in February 2026 on a Consultation Draft version of this report that was made available in December 2025. The authors greatly appreciate the time stakeholders have given to support the project. Submissions received can be viewed at https://www.aemo.com.au/consultations/current-and-closed-consultations/draft-2026-forecasting-assumptions-update.
GenCost 2025-26 | ix
Executive summary
Technological change in electricity generation is a global effort that is strongly linked to global climate change policy ambitions. While the rate of change remains uncertain and the level of commitment of each country varies over time, in broad terms, there is continued support for collective action limiting global average temperature increases. At a domestic level, the Commonwealth government, together with Australian states and territories aspire to or have legislated net zero emissions (NZE) by 2050 targets.
Globally, renewables (led by wind and solar PV) are the fastest growing energy source, and the role of electricity is expected to increase materially over the next 30 years with electricity technologies presenting some of the lowest cost abatement opportunities.
Purpose and scope
GenCost is a collaboration between CSIRO and AEMO to deliver an annual process of updating the capital costs of electricity generation, energy storage and hydrogen production technologies with a strong emphasis on stakeholder engagement. GenCost represents Australia’s most comprehensive electricity generation cost projection report. It uses the best available information each cycle to provide an objective annual benchmark on cost projections and updates forecasts accordingly to guide decision making, given technology costs change each year. This is the eighth update following the inaugural report in 2018.
Technology costs are one piece of the puzzle. They are an important input to electricity sector analysis which is why we have made consultation an important part of the process of updating data and projections.
The report encompasses updated current capital cost estimates commissioned by AEMO and delivered by GHD. Based on these updated current capital costs, the report provides projections of future changes in costs consistent with updated global electricity scenarios which incorporate different levels of achievement of global climate policy ambition. Levelised costs of electricity (LCOEs) are also included and provide a summary of the relative competitiveness of generation technologies.
Responses to stakeholder feedback
The GenCost consultation draft introduced a major new feature which was the system levelised cost of electricity (SLCOE) which provides a more transparent way of capturing all of the system wide costs that are needed to support the increasing share of renewables that are being deployed in Australia’s electricity system. These broader costs are unable to be captured in the standard levelised cost of electricity (LCOE) which is applied on a single technology basis. The new SLCOE method also included the publication of a new open source electricity system model and input data set.
Stakeholders generally supported the new method but were less convinced of the value of the SLCOE in measuring cost in 2030 than in 2050. In 2030, cost estimates are strongly influenced by existing capacity and there is no consistent rule or readily available data for capturing the amount
x | CSIRO Australia’s National Science Agency
of unrecovered costs associated with that capacity. In response to this feedback, for 2030 SLCOE estimates have been replaced with electricity futures prices as an alternative indicator of generation prices in the period to 2030. These rely on market expectations rather than modelling and are publicly available.
Stakeholders also requested more detail on SLCOE modelling results. A detailed set of results for the 2050 SLCOE modelling is now included in the GenCost data file that accompanies this report. Other minor points of feedback and responses are summarised in the body of the report and submissions can be viewed at AEMO’s consultation website.
Response to the 2026 Iran War
The Iran war began on 28 February 2026 constricting global oil products supply. The depth and direction of its broader impacts for energy technologies is unpredictable. It is inflationary in the sense of oil being a fundamental input to almost all global supply chains. It may also increase demand for energy technologies in general that reduce a region’s reliance on imported oil. It is also deflationary because it could reduce global economic growth and demand for energy. In light of this development, for 2026, we have assumed no improvement in costs for the majority of technologies.
Data centre impacts on gas turbine costs
New electricity generation capacity required to meet demand from data centres is growing strongly and is impacting gas turbine costs. The IEA (2026) reported that:
“if data centres were a country, they would be the second-largest destination for gas turbines ordered from Q1 2025 to Q1 2026”.
Gas turbine costs have already increased for the last four consecutive years. As a result of this ongoing demand from data centres, the projections assume that the cost of gas turbine technologies will continue to increase in the next two years after which manufacturers are expected to be in a better position to meet demand, allowing costs to stabilise.
2030 costs
Baseload electricity futures prices are in the range of $68/MWh to $92/MWh for the next 3 years with a flat or rising trend, depending on the state. Peak evening futures prices range from $152/MWh to $220/MWh.
Peaking generation technologies are currently sitting at the high end of Peak evening futures prices. This is consistent with observations that batteries have started to compete with traditional gas generation in this category, with lower battery costs and significant new capacity added resulting in reduced peak evening prices.
Similarly, most bulk energy generation technologies are higher cost than the Baseload electricity futures price. This is an academic observation because most of these technologies could not be deployed within the next 4 years due to their required development and construction times.
Solar PV and onshore wind are almost exclusively the only generation technologies sufficiently advanced in the development pipeline to be deployed before 2030. Their deployment is being supported by existing flexible gas and coal together with new battery, pumped hydro and
GenCost 2025-26 | xi
transmission capacity. Given the Baseload electricity futures price is an all-day, all-year price it represents the expected total annual generation cost of this combination of technologies. Transmission costs, which are recovered outside of the electricity system increase to 2030 but from a low base (AEMC 2025).
LCOE does not include renewable integration costs.
ES Figure 0-1 Levelised cost of electricity (LCOE) in 2030 by technology category and corresponding electricity futures price
2050 generation technology competitiveness
The key findings from the 2050 analysis are:
•
For the electricity sector to efficiently support achieving net zero by 2050 across the economy, solar PV and onshore wind will deliver the majority of electricity supply (93%) supported by hydro, storage, transmission and either gas or hydrogen or a combination of both.
•
New black coal, while competitive at this SLCOE range is not suitable for deployment if the goal is to efficiently achieve net zero (coal would increase the cost of achieving net zero across the economy). Unlike coal, the SLCOE analysis indicates gas will play a role in achieving net zero emissions contributing a 3% to 7% share of generation.
•
Solar thermal is competitive relative to other technologies and inside the SLCOE range. However, given the need to access better solar resources which are further from load centres, solar thermal will be subject to additional transmission costs of up to $20/MWh.
•
Some offshore wind is also in the competitive range however the modelling used the average cost of offshore wind. Lower range offshore wind costs appear to be competitive but were not explored in the modelling and so need more analysis.
01002003004005006007002025 $/MWh2030 LCOE rangeFutures price lowFutures price highPeaking technologies Bulk energy technologies Variable technologies
xii | CSIRO Australia’s National Science Agency
•
Gas with CCS is the next most competitive after solar thermal and offshore wind. Large-scale nuclear is slightly higher in cost than gas with CCS. Black coal with CCS occupies a similar cost range to large-scale nuclear. Nuclear small modular reactors (SMRs) are the highest cost.
LCOE does not include renewable integration costs. SLCOE does.
ES Figure 0-2 Levelised cost of electricity (LCOE) in 2050 by technology and system levelised cost of electricity (SLCOE) for efficiently achieving net zero by 2050
Implications for electricity prices
In 2025 calendar year the average NEM volume weighted generation price was estimated to be $104/MWh, down from the recent peak in 2022 of $189/MWh caused predominantly by high gas prices. Based on NEM electricity futures prices in the next three years (and supported by AEMC (2025) modelling), generation costs to 2030 are expected to fall further relative to 2025, down to $80-$90/MWh.
By 2050, most generation capacity that exists today will be retired. The LCOE and SLCOE analysis in this report indicates that there are no new build technology options, fossil or non-fossil with firming costs, that cost less than $100/MWh. By 2050, new build coal generation costs are in the range of $107/MWh to $182/MWh. As such, even without net zero by 2050 policies, generation prices are expected to increase above $100/MWh in the post-2030 period.
The cost of generation if the electricity sector makes an efficient contribution to achieving net zero by 2050 is $141/MWh to $152/MWh based on solar PV and wind generation inclusive of all firming costs. Excluding transmission costs which are not recovered through the generation market, the net zero consistent generation cost is $120/MWh to $130/MWh or an average of $125/MWh.
These cost estimates do not guarantee future generation prices. Changes in generation prices are also subject to:
01002003004002025 $/MWhLCOE rangeSLCOE lowSLCOE high
GenCost 2025-26 | xiii
•
Supply-demand imbalance as a result of too much or too little deployment relative to demand growth and retirements.
•
Fuel price and weather volatility.
•
The level of competition amongst suppliers.
These additional drivers of generation price formation can lead to prices significantly lower or higher than the underlying cost of the system and can take many years to correct due to the long lead times for capacity deployment. Generation prices are currently around 33% of retail prices. Transmission is around 7%, distribution around 34% with the remainder made up of metering, retail services and government programs.
14 | CSIRO Australia’s National Science Agency
1 Introduction
Current and projected electricity generation, storage and hydrogen technology costs are a necessary and highly impactful input into electricity market modelling studies. Modelling studies are conducted by the Australian Energy Market Operator (AEMO) for planning and forecasting purposes. They are also widely used by electricity market actors to support the case for investment in new projects or to manage future electricity costs. Governments and regulators require modelling studies to assess alternative policies and regulations. There are substantial coordination benefits if all parties are using similar cost data sets for these activities or at least have a common reference point for differences.
The report provides an overview of updates to current costs in Section 2. This section draws significantly on updates to current costs provided in GHD (2026) and further information can be found in their report. The global scenario narratives are outlined in Section 3. Capital cost projection results are reported in Section 4 and LCOE and SLCOE results in Section 5. CSIRO’s cost projection methodology is discussed in Appendix A including key global data assumptions. Appendix B provides data tables and this data can also be downloaded from CSIRO’s Data Access Portal1. A set of technology selection and data quality principles has been included in Appendix E. Feedback on these principles is always welcome.
1.1 Scope of the GenCost project and reporting
The GenCost project is a joint initiative of the CSIRO and AEMO to provide an annual process for updating electricity generation, storage and hydrogen technology cost data for Australia. The project is committed to a high degree of stakeholder engagement as a means of supporting the quality and relevancy of outputs. Each year a consultation draft is released in December for feedback before the final report is completed towards the end of the financial year.
The project is flexible about including new technologies of interest or, in some cases, not updating information about some technologies where there is no reason to expect any change, or if their applicability is limited. Appendix E discusses some technology inclusion principles. GenCost does not seek to describe the set of electricity generation and storage technologies included in detail. The report provided by GHD (2026) does include more detailed technology specifications and commentary.
1.1.1 CSIRO and AEMO roles
AEMO and CSIRO jointly fund the GenCost project by combining their own resources. AEMO commissioned GHD to provide an update of the current cost and performance characteristics of electricity generation, storage and hydrogen technologies (GHD, 2025). This report focusses on
1 Search GenCost at https://data.csiro.au/collections
GenCost 2025-26 | 15
capital costs, but the GHD report provides a wider variety of data such as operating and maintenance costs and energy efficiency. Some of these other data types are used in levelised cost of electricity calculations in Section 5.
Project management, capital cost projections (presented in Section 4), LCOE estimates (Section 5) and development of this report are primarily the responsibility of CSIRO.
1.1.2 Incremental improvement and focus areas
There are many assumptions, scope and methodological considerations underlying electricity generation and storage technology cost data. In any given year, we are readily able to change assumptions in response to stakeholder input. However, the scope and methods may take more time to change, and input of this nature may only be addressed incrementally over several years, depending on the priority.
In this report, the main innovation is an update of the method for calculating technology integration costs. The method used up until now was designed in 2018 and its revision and new results are discussed in Section 5. The new method accounts for feedback we received about the previous method and the development of the new method has been separately published in Graham et al. (2025).
1.2 The GenCost mailing list
The GenCost project would not be possible without the input of stakeholders. No single person or organisation can follow the evolution of all technologies in detail. We rely on the collective deep expertise of the energy community to review our work before publication to improve its quality. To that end the project maintains a mailing list to share draft outputs with interested parties. The mailing list is open to all. To join, use the contact details on the back of this report to request your inclusion. Some draft GenCost outputs are also circulated via AEMO’s Forecasting Reference Group mailing list which is also open to join via their website.
1.3 Summary of feedback received and changes to the consultation draft
1.3.1 Technology specific items
Solar thermal
A new report on the economics of Australian solar thermal by Fichtner (2026) became available after the release of the consultation draft. Consequently, several of the cost and performance factors for solar thermal have been updated to be consistent with this new information.
Pumped hydro energy storage (PHES)
PHES current capital costs have increased relative to the draft report in recognition of additional data becoming available regarding the costs of a current PHES development. The current
16 | CSIRO Australia’s National Science Agency
development provides evidence of higher costs and has been incorporated into the new estimates by GHD (2026).
Offshore wind
The economic life of offshore wind has been increased from 25 to 30 years to be consistent with international approaches. In the LCOE calculations, the maximum capacity factor for off-shore wind in 2050 has been reduced from 61% to 55%. This change makes the capacity factor assumption more consistent with fixed off-shore wind projects. Higher capacity factors could be achieved with floating offshore wind but that technology is not currently included in the LCOE analysis. These changes do not impact the SLCOE analysis which directly uses renewable production profiles for each relevant NEM renewable energy zone (rather than simple average capacity factors) and includes both fixed and floating offshore wind options.
Nuclear
Current capital costs for nuclear SMR continue to be based on the Darlington project in Canada. The Australian interpretation of this capital cost data has been revised downwards to recognise that the published cost for that project includes interest during construction. Capital costs in GenCost do not include that factor but rather are expressed as an ‘overnight’ costs. For all technologies, interest during construction is added back in when converting capital costs to levelised costs of electricity (LCOE).
In the system levelised cost of electricity analysis, the 2050 least cost generation mix is calculated on the basis of an average emissions intensity level consistent with the electricity sector cost-effectively contributing to a national net zero target. Nuclear generation was a assigned a zero emission intensity in the draft report. However, it now has a small positive emissions intensity associated with its scope 3 emissions associated with nuclear fuel processing, extraction and transport. This change was to be consistent with coal and gas whose emissions factors also include these scope 3 emissions associated with fuel supply.
Electrolysis
There continue to be diverse views on the current capital cost of electrolysis projects. These arise mainly from differences in emphasis on available project examples and different definitions of project boundaries. The approach in the GHD (2026) report remains unchanged and includes detail on project boundaries applied.
This final GenCost report updated to the 2025 IEA World Energy Outlook. This included a significant reduction in long term hydrogen demand under the Global NZE post 2050 scenario compared to 2024. There was a smaller decrease in hydrogen demand under the Current Policies and Global NZE by 2050 scenarios. Constraints were also placed on the maximum build out of electrolysers in Australia based on market constraints and are consistent with AEMO scenarios. Given costs are reduced through deployment, a direct consequence of lower hydrogen demand is that electrolyser cost reductions are more limited in the updated projection.
Solar PV
Differences in solar PV learning rates between scenarios which are used to project future changes in capital costs have been simplified for greater consistency.
GenCost 2025-26 | 17
1.3.2 General items
Consistency of approach to land and development costs
There was a request for consistency to be applied across some technologies in the approach to calculating the land and development costs in current capital costs. This request was not supported on the basis that there are important differences between technologies that warrant a technology specific approach.
Treatment of historical investments in 2030 SLCOE
The primary goal of providing LCOE information is so that anyone can quickly compare the cost of different technologies. As many have noted, this approach is not directly possible with renewable generation technologies because they require a portfolio of technologies to provide reliable power. Hence, for renewables GenCost calculates the cost of a full system dominated by renewable energy but including additional costs such as transmission, storage and peaking generation. With this approach the average cost of such a new system can be compared to the cost of new builds of technologies such as gas, coal and nuclear. This information tells us two useful things:
•
- Whether renewables can compete with other new technologies and vice versa
•
- The likely cost of electricity generation for renewable dominated systems in the future
This system costing approach is well-accepted when examining a far-off year such as 2050 when we can assume most existing capacity has retired and is therefore not included as either a resource or a cost. However, many stakeholders remain uncomfortable in using this approach for near term analysis such as calculating 2030 system costs. The issue that arises is what should be done with costing existing capacity? There will be existing capacity in the system which cannot be ignored. Some of the cost of existing capacity has already been recovered, some will be recovered in the future (particularly transmission costs which are recovered through regulatory arrangements) and some costs may either never be recovered or could be more than recovered (i.e. producing profits above planned rates of return) depending on future market conditions and individual contracting and hedging arrangements.
In the 2025-26 consultation draft, GenCost assigned no cost to existing or under construction capacity in 2030 which is a fairly typical approach in economic modelling. However, despite this deliberate omission, the 2030 cost of generation projected by the system cost modelling, if applied as the average generation price, would likely allow for reasonable recovery of past generation investment costs as well. This is perhaps surprising given no cost was assigned to existing capacity. However, so long as new build costs are not too different from past costs then the cost of new build can be similar to the market price needed for existing market participants to recover their costs.
Whilst our approach to calculating 2030 generation costs is indicative of an electricity price that would allow for both existing and new generation capacity to recover costs, some stakeholders will likely remain uncomfortable with the approach as it does not specifically explore the costs of those recent historical investments. For this reason, we have decided to find an alternative measure of 2030 generation costs.
18 | CSIRO Australia’s National Science Agency
The new source of generation costs to 2030 is the NEM electricity futures prices (volume weighted across the NEM states excluding Tasmania). Futures prices provide buyers with a hedge against higher prices at the potential cost of not benefiting from lower price outcomes should they occur. For sellers, a futures contract can be used to protect them against lower price outcomes at the cost of not gaining from higher priced outcomes, should they occur. With this context, a supplier is only motivated to contract if the futures contract price already covers all of their costs since they will not receive any additional revenue on the contracted volume. Therefore, the futures price is a good indicator of expected generation costs of suppliers in that future year. The futures price we use to compare to bulk energy technologies is the Baseload price for all hours of a full year and so has the same time scope as the average system price previously calculated through SLCOE modelling. There is also a peak evening futures price that is relevant for peaking technologies.
The change in method for costing the system in 2030 has not led to a significant change in results. For now, the Baseload futures prices is reasonably well aligned with the consultation draft 2030 SLCOE estimate.
SLCOE model
The following improvements and upgrades were made to the Simple Electricity Model:
•
All input data has been updated to be consistent with either
–
Draft 2026 ISP Input and Assumptions Workbook, or
–
Draft 2026 Forecasting Assumptions Update Workbook, and
–
This version of GenCost
Data from the Final 2026 ISP could not be included as its release date did not allow enough time for it to be considered for this publication.
•
The updated data resulted in two additional weather years being available, however, after testing, 2011 remains the highest cost weather year among those available and so is the year use in the SLCOE modelling.
•
Nuclear now has an emissions factor to be consistent with the gas and coal technologies which recognise upstream emissions from fuel supply. As a consequence, the previous Zero Emission scenario included in the consultation draft that had a target of 0tCO2e/MWh has been replaced with a scenario called VeryStrongNetZero with an emission target of 0.005tCO2e/MWh. While not zero, it only allows for about 1 million tonnes of CO2e in 2050.
Cost year
For greater clarity all costs charts now refer to the calendar year rather than financial year. See the beginning of Appendix B for a longer discussion of the cost year basis.
GenCost 2025-26 | 19
2 Current technology costs
2.1 Current cost definition
Our definition of current capital costs is current contracting costs or costs that have been demonstrated to have been incurred for projects completed in the current financial year (or within a reasonable period before). We do not include in our definition of current costs, costs that represent quotes for potential projects or project announcements.
While all data is useful in its own context, our approach reflects the objective that the data must be suitable for input into electricity models. The way most electricity models work is that investment costs are incurred either before (depending on construction time assumptions) or in the same year as a project is available to be counted as a new addition to installed capacity2. Hence, current costs and costs in any given year must reflect the costs of projects completed or contracted in that year. Quotes received now for projects without a contracted delivery date are only relevant for future years. This point is particularly relevant for technologies with fast-reducing costs. In these cases, lower cost quotes will become known in advance of those costs being reflected in recently completed deployments – such quotes should not be compared with current costs in this report but with future projections.
For technologies that are not frequently being constructed, our approach is to look overseas at the most recent projects constructed. This introduces several issues in terms of different construction standards and engineering labour costs which have been addressed by GHD (2026). GHD (2026) also provide more detail on specific definitions of the scope of cost categories included. GHD cost estimates are provided for Australia in Australian dollars. They represent the capital costs for a location not greater than 200km from the Victorian metropolitan area. GHD provide adjustments for costs for different regions of the NEM. Site conditions will also impact costs to varying degrees, depending on the technology. CSIRO adjusts the data when used in global modelling to take account of differences in costs in different global regions.
GHD (2026) also provides detailed information on the boundary of capital costs such as what development costs are included, ambient temperature, distance to fuel source and many other considerations.
2.1.1 First-of-a-kind cost premiums
When building a technology that has a degree of novelty, capital cost estimates typically underestimate the realised cost of installation. This is sometimes called an optimism factor or first-of-a-kind (FOAK) costs. These costs are reduced with more installations. The industry term for the point when costs are no longer impacted by the immaturity of the development supply chain is
2 This is not strictly true of all models but is most true of long-term investment models. In other models, investment costs are converted to an annuity (adjusted for different economic lifetimes), or additional capital costs may be added later in a project timeline for replacement of key components.
20 | CSIRO Australia’s National Science Agency
nth-of-a-kind (NOAK). The cost estimates in GenCost are mostly on a NOAK basis. This is not because all technologies have mature supply chains but rather because it is too difficult to objectively estimate the FOAK premium that should be applied. It is only observable after a proponent fails to deliver the first project for the cost they had planned. Even then it is difficult to separate optimism from ordinary changes in circumstances, particularly for projects that have long total development times. These cost increases will sometimes be found through the process of more detailed engineering and feasibility studies prior to final investment decisions but may not be shared publicly.
EIA (2023) applies FOAK premiums of up to 25% to their technology costs. AACE (1991) recommends applying different levels of contingency based on the Technology Readiness Level ranging from 10% to up to 70%. In practice, we can find examples of projects that have cost around 100% more than planned such as the Vogtle large-scale nuclear plant in the US and the Snowy 2.0 pumped hydro project in Australia. Flyvbjerg and Gardner (2023) report that the global average cost overrun for nuclear, hydro, wind and solar are 120%, 75%, 13% and 1%, respectively. As such, while special circumstances may have occurred in specific cases, generally, FOAK premiums should be part of normal expectations for estimating the cost of deploying less mature or large technology projects in the future.
The technologies most at risk of FOAK cost premiums in Australia are:
•
Offshore wind
•
Large-scale nuclear
•
Small modular reactor (SMR) nuclear
•
Solar thermal
•
Coal, gas or biomass with carbon capture and storage
•
Wave, tidal and ocean current technologies.
Given the size and unique site conditions of most pumped hydro projects they may also continue to be at risk of cost overruns. However, given these projects are relatively rare, in practice there is not as much difference between a FOAK and NOAK costing.
Technologies that are currently being regularly deployed in Australia such as onshore wind, solar PV, batteries and gas generation are least likely to be impacted. Technologies that have been deployed before and are globally commercially mature may still be subject to FOAK premiums due to large intervals since the last deployment leading to loss of skills, new designs which create uncertainty or new licensing requirements, project size and unique site conditions.
It is likely that 2024 nuclear SMR costs included some FOAK costs given it was based on a FOAK in the US project. However, the first commercial project is proceeding in Canada at Darlington and costs are reduced from the 2024 level to match that project’s costings which include the assumption that they will build each of the four proposed units at lower cost than the previous unit. Regardless of how successful Canada is in reducing costs for each unit build, Australia would still experience a FOAK premium if that technology were to be built for the first time here.
FOAK premiums are very difficult to forecast. However, given stakeholder interest in the technologies listed above, there is a need for an estimate of the FOAK premium (Table 2-1). To develop the premium the value of 120% has been applied to large scale nuclear based on Flyvbjerg
GenCost 2025-26 | 21
and Gardner (2023). The remaining premiums are based on observing the ratio between this large scale nuclear premium and its construction time and applying that ratio to the other technology’s construction times. Effectively we are proposing that technologies that take longer to build will face higher FOAK premiums as they are more complex to plan. We then halve the premium for the second project and assume the third and subsequent projects are not impacted by a FOAK premium.
Table 2-1 Suggested FOAK premium by technology
Construction time Premium Technology
Years
First project
Second project Gas with CCS
2.0
42%
21% Black coal with CCS
2.0
42%
21% Nuclear SMR
4.4
92%
46% Nuclear large-scale
5.8
120%
60% Solar thermal
1.8
37%
18% Wind offshore
3.0
63%
31%
2.2 Capital cost source
AEMO commissioned GHD (2026) to provide an update of current cost and performance data for existing and selected new electricity generation, storage and hydrogen production technologies. We have used data supplied by GHD (2026) which represents a July estimate and so it is consistent with either the beginning of the financial year 2025-26 or the middle of 2025.
Nuclear technologies are not included in GHD (2026). These are sourced separately by CSIRO.
2.3 Current generation technology capital costs
Figure 2-1 provides capital costs for selected technologies since the project’s inception in 2018. All costs are expressed in real 2025-26 Australian dollars, represent overnight costs and do not include any available subsidies.
Costs increased for many technologies from 2022 owing to the global supply chain constraints following the COVID-19 pandemic which also increased freight and raw material costs. Technologies were impacted differently given different input materials and manufacturing regions and are recovering from this development at different rates. The change in current costs over the past four years indicates an easing of inflationary pressures for solar PV, wind and batteries while coal and gas technology costs have recently increased significantly (Figure 2-2). Coal and gas costs are not developed from project data but rather using standard industry software since there are insufficient projects in Australia. In particular, existing coal projects are very old and therefore unreliable for estimating current costs. The latest updates to industry software included a large upwards revision in gas turbine and steam turbine costs.
22 | CSIRO Australia’s National Science Agency
Figure 2-1 Comparison of current capital cost estimates with previous reports (in real terms)
Figure 2-2 Year on year change in current capital costs of selected technologies in the past four years (in real terms)
0
2000
4000
6000
8000
10000
12000
14000
Black coal Black coal
with CCS
Gas
combined
cycle
Gas open
cycle (large)
Gas with CCS Large scale
solar PV
Solar thermal
(14hrs)
Wind
2025 $/kW
2018 2019 2020 2021 2022 2023 2024 2025
17%
13%
16%
9%
35%
20%
-2%
14%
6%
-8%
8%
4% 2%
11%
19%
-8%
6%
-20%
13%
0%
32%
9%
-5%
-15%
Black coal Gas combined
cycle
Gas open cycle
(large)
Large scale solar
PV
Wind (onshore) Large scale
battery (2hr)
2022-23 2023-24 2024-25 2025-26
GenCost 2025-26 | 23
2.4 Current storage technology capital costs
Updated and previous capital costs are provided on a total cost basis for various durations3 of batteries and pumped hydro energy storage (PHES) in $/kW and $/kWh. GenCost only provides projections for batteries and PHES. Current costs of compressed air energy storage are included in GHD (2026).
None of these capital costs provide enough information to be able to say one technology is more competitive than the other. Capital costs are only one factor. Additional cost factors include energy input costs (where not already included), utilisation rate, round trip efficiency, operating costs and design life.
Total cost basis means that the costs are calculated by taking the total project costs divided by the capacity in kW or kWh4. As the storage duration of a project increases then more batteries or larger reservoirs need to be included in the project, but the power components of the storage technology remain constant. As a result, $/kWh costs tend to fall with increasing storage duration (Figure 2-3).
Note that these $/kWh costs are not for energy delivered but rather a capacity of storage. GenCost does not present levelised costs of storage (LCOS) which are on an energy delivered basis. However, LCOS estimates are available from the CSIRO (2023) Renewable Energy Storage Roadmap.
Storage capital costs in $/kW increase as storage duration increases because additional storage duration adds costs without adding any additional power capacity to the project (Figure 2-4). Additional storage duration is most costly for batteries. These trends are one of the reasons why batteries tend to be deployed in low storage duration applications, while PHES is deployed in high duration applications. A combination of durations may be required by the system depending on the operation of other generation in the system, particularly the scale of variable renewable generation and peaking plant (see Section 5).
Depth of discharge in batteries can be an important constraint on use. However, all GHD battery costs are presented on a usable capacity basis such that the depth of discharge is 100%5. GHD (2026) also includes estimates of battery costs when they are integrated within an existing power plant and can share some of the power conversion technology. This results in around a 5% lower battery cost for a 1-hour duration battery, scaling down to a 2% cost reduction for 8 hours duration. PHES is more difficult to co-locate.
3 The storage duration used throughout this report refers to the maximum duration for which the storage technology can discharge at maximum rated power. However, it is important to note that every storage technology can discharge for longer by doing so at a rate lower than their maximum rated power
4 Component costs basis is when the power and storage components are separately costed and must be added together to calculate the total project cost.
5 The batteries in this publication have additional capacity which is not usable (e.g., there is typically a minimum 20% state of charge). This unusable capacity is not counted in the capacity of the battery or in any expression of its costs. When other publications include this unusable capacity the depth of discharge is less than 100%.
24 | CSIRO Australia’s National Science Agency
Figure 2-3 Capital costs of storage technologies in $/kWh (total cost basis)
Battery costs (battery and balance of plant in total) have decreased significantly by 11% to 16% depending on the duration.PHES current cost estimates have decreased in 2025. See GHD (2026) for details on how the costs were prepared. It is important to note that PHES has a wider range of uncertainty owing to the greater influence of site-specific issues. Batteries are more modular and as such costs are relatively independent of the site.
Concentrating solar thermal (CST) is another technology incorporating storage but it is reported as a generation technology in Section 5. It incorporates built-in long-duration energy storage. Direct comparison with the other electricity storage technologies is complicated by the fact that a CST system also collects its own solar energy. Direct comparison with other storage technologies via calculation of the LCOS can be found in CSIRO’s Renewable Energy Storage Roadmap (CSIRO, 2023), but is outside the scope of GenCost.
0200400600800100012002025 $/kWh2019202020212022202320242025
GenCost 2025-26 | 25
Figure 2-4 Capital costs of storage technologies in $/kW (total cost basis)
0200040006000800010000120002025 $/kW2019202020212022202320242025
26 | CSIRO Australia’s National Science Agency
3 Scenario narratives and data assumptions
The global scenario narratives for Current Policies and Global NZE by 2050 included in GenCost have not changed since GenCost 2022-23 but there have been some updates to data assumptions. The narrative of Global NZE post 2050 has changed slightly to reflect changes in the matching IEA scenario.
3.1 Scenario narratives
The global climate policy ambitions for the Current policies, Global NZE post 2050 and Global NZE by 2050 scenarios have been adopted from the International Energy Agency’s (IEA) 2025 World Energy Outlook (IEA, 2025) scenario matching to the Current Policies scenario, Stated Policies (STEPS) scenario, and Net Zero Emissions by 2050. Previous versions of the World Energy Outlook included an Announced Pledges Scenario, which was matched to GenCost’s Global NZE post 2050 scenario and the STEPS scenario to GenCost’s Current policies scenario. The IEA has adjusted its scenario definitions to be less ambitious, except for the Net Zero Emissions by 2050 scenario.
Various elements, such as the degree of vehicle electrification and hydrogen production, are also consistent with the matching IEA scenarios.
3.1.1 Current policies
The Current policies scenario includes existing climate policies as at mid-2025 and does not assume that all government targets will be met. The implementation of climate policies in the modelling includes a combination of carbon prices and other climate policies6. This scenario has the strongest constraints applied with respect to global variable renewable energy resources and the slowest technology learning rates. This is consistent with a lack of any further progress on emissions abatement beyond recent commitments. Demand growth is moderate with moderate electrification of transport and limited hydrogen production and utilisation. Total greenhouse gas emissions lead to a global average surface temperature rise of around 2 °C in 2050 and 2.9 °C in 2100.
3.1.2 Global NZE post 2050
The Global NZE post 2050 has moderate renewable energy constraints and middle-of-the-range learning rates. It includes existing policies, a carbon price and policy announcements that are consistent with the broad direction the energy sector is travelling in. It does not assume that
6 The application of a combination of carbon prices and other non-carbon price policies is consistent with the approach applied by the IEA. While we directly apply the IEAs published carbon prices, we design our own implementation of non-carbon price policies to ensure we match the emissions outcomes in the relevant IEA scenario. Structural differences between GALLM and the IEA’s models means that we cannot implement the exact same non-carbon price policies. Even if our models were the same, the IEA’s description of non-carbon price policies is insufficiently detailed to apply directly.
GenCost 2025-26 | 27
governments will meet their Nationally Determined Contributions (NDCs) and longer-term net zero emission targets. Hydrogen trade and transport and industry electrification are similar to Current policies. By 2100, the global temperature is projected to rise by 2.5 °C
3.1.3 Global NZE by 2050
Under the Global NZE by 2050 scenario there is a strong climate policy consistent with achieving net zero emissions by 2050 worldwide. The achievement of these abatement outcomes is supported by the strongest technology learning rates and the least constrained (physically and socially) access to variable renewable energy resources. Reflecting the low emission intensity of the predominantly renewable electricity supply, there is an emphasis on high electrification across sectors such as transport, hydrogen-based industries and buildings leading to the highest electricity demand across the scenarios. This scenario reflects the fact that exceeding 1.5 °C is now inevitable. However, by targeting net zero by 2050 and with further actions beyond 2050, the global average temperature increase falls back below 1.5 °C by 2100.
Table 3-1 Summary of scenarios and their key assumptions Key drivers Current policies Global NZE post 2050 Global NZE by 2050 IEA WEO scenario alignment Current policies scenario Stated policies scenario Net zero emission by 2050
CO2 pricing / climate policy
Based on current policies only
Based on current policies and policies consistent with the direction of travel of the energy sector
Consistent with 1.5 degrees world in 2100. Renewable energy targets and forced builds / accelerated retirement Current renewable energy policies Renewable energy policies extended as needed High reflecting confidence in renewable energy
Demand / Electrification
Medium
Medium
High Learning rates1 Weaker Normal maturity path Stronger
Renewable resource & other renewable constraints2
More constrained than existing assumptions
Existing constraint assumptions
Less constrained
1 The learning rate is the potential change in costs for each doubling of cumulative deployment, not the rate of change in costs over time. See Appendix C for assumed learning rates.
2 Existing large-scale and rooftop solar PV renewable generation constraints are as shown in Apx Table C.4.
28 | CSIRO Australia’s National Science Agency
4 Projection results
All projections start from a current cost observed in 2025. The primary source of 2025 costs is GHD (2026) with data gathered from other sources where otherwise not available in that report. All projections are in real terms. That is, all projected cost changes after 2025 are in addition to the general level of inflation.
4.1 Short-term and long-term inflationary pressures
4.1.1 Short term equipment costs
In recent years, the cost of a range of technologies including electricity generation, storage and hydrogen technologies has increased rapidly driven by two key factors: increased freight and raw materials costs. The most recent period where similar large electricity generation technology cost increases occurred was 2006 to 2009 with wind turbines and solar PV modules being most impacted. The cost drivers at that period of time were policies favouring renewable energy in Europe, which led to a large increase in demand for wind and solar. This coincided with a lack of supply due to insufficient manufacturing facilities of equipment and polysilicon in the case of PV and profiteering by wind turbine manufacturers (Hayward and Graham, 2011). Once supply caught up with demand, the costs returned to a trajectory consistent with learning-by-doing and economies of scale.
CSIRO has explored a number of resources to understand cost increases already embedded in technology costs and to project how this recent increase in costs will resolve. We normally use our model GALLM to project all costs from the current year onwards. While GALLM takes into account price bubbles caused by excessive demand for a technology (as happened in 2006-2009), the drivers of the current situation are different and thus we have decided to take a different approach, at least for projecting costs over the next decade. It is not appropriate to project long-term future costs directly from the top of a price bubble, otherwise all future costs will permanently embed what may be temporary market features.
It is acknowledged that some stakeholders believe the price bubble is not temporary but rather a permanent feature of future costs. However, to sustain real price increases, supply needs to be either constrained by a cartel (or other persistent market power) or resource scarcity or technology demand needs to grow faster than supply (which implies strong non-linear demand growth since, once established, a given supply capacity can meet linear growth at the rate of that existing capacity7). The current cost update indicates inflationary pressures are weakening for at least some technologies.
7 If the world ramps up to X GW per year technology manufacturing capacity by a certain date, then, without expanding manufacturing capacity any further, it can meet any future capacity target after that date up to the value of bX (where b is the years since the manufacturing capacity was established). The future capacity target would need to include all capacity needed to meet growth as well as replace retiring plant.
GenCost 2025-26 | 29
Historical experience and the projections available for global clean energy technology deployment do not provide confidence that the required market circumstances for sustained real price increases will prevail for the entire projection period (see Appendix D of the GenCost 2022-23: Final report for more discussion on this topic). However, it is considered that the period to 2030 will likely experience extra strong technology deployment. This is partly because of the low global clean technology base (from which non-linear growth is more feasible) but also because governments and industry often use the turning of a decade as a target date for achieving energy targets.
Our current view is that it may take longer than 2030 for technologies to return to a more normal level of costs. This report assumes that if a technology has not already started to show strong signs of recovery it will not return to their normal cost path until 2035. This includes technologies such as onshore wind, coal, gas and nuclear. Note that to achieve a recovery in technology costs they need only stay constant in nominal prices. This delivers real cost reduction of around 2-3% per year.
A consequence of this modelling approach is that the near-term cost reductions that are shown in the following pages mostly do not reflect learning. Rather, they are predominantly the slow unwinding of inflationary pressures that have temporarily placed costs above the underlying cost curve. Solar PV, batteries, fuel cells and offshore wind have already passed through the global inflationary event and their costs now follow the underlying learning curve cost trajectory.
Response to the 2026 Iran war
The Iran war began on 28 February 2026 constricting global oil products supply. The depth and direction of the broader impacts for energy technologies is unpredictable. It is inflationary in the sense of oil being a fundamental input to almost all global supply chains. It may also increase demand for energy technologies in general that reduce a region’s reliance on imported oil. It is also deflationary because it could reduce global economic growth and demand for energy. In light of this development, for 2026, we have assumed no improvement in costs for the majority of technologies. Solar PV and batteries are excluded as they have previously demonstrated greater resilience against global inflationary pressures.
Data centre impacts on gas turbine costs
New electricity generation capacity required to meet demand from data centres is growing strongly and is impacting gas turbine costs. The IEA (2026) reported that:
“if data centres were a country, they would be the second-largest destination for gas turbines ordered from Q1 2025 to Q1 2026”.
Gas turbine costs have already increased for the last four consecutive years. As a result of this ongoing demand from data centres, the projections assume that the cost of gas turbine technologies will continue to increase in the next two years after which manufacturers are expected to be in a better position to meet demand, allowing costs to stabilise.
30 | CSIRO Australia’s National Science Agency
4.1.2 Long term land and construction costs
Two exceptions where scarcity is a factor and is expected to lead to ongoing real increases in costs is land and construction costs. Land costs generally make up 2% to 9% of generation, storage and electrolyser capital costs. The projections take the land share of capital costs provided in GHD (2026) and inflate that proportion of costs by the real land cost index that is published in Mott MacDonald (2023)8. This common land cost index provides some consistency between the treatment of land costs between transmission, generation and storage assets in AEMO’s modelling. The inclusion of a specific land cost inflator was first included in the GenCost 2022-23: Final report.
Information on future real construction costs become available in a February 2025 report from Oxford Economics Australia (2025), commissioned by AEMO. The data indicates that while construction costs are expected to ease in the short term, a longer-term trend of rising real construction costs is projected owing primarily to above inflation growth in construction workers’ wages and, to a lesser extent, constrained supply of quarry and cement materials. The construction cost escalation factors estimated by Oxford Economics Australia are applied to the installation cost proportion of capital costs which is sourced from GHD (2026). Note that, this escalation factor is applied after learning. That is, it is still possible for developers to be more productive or innovative at installing some technologies while at the same time facing increases in real costs for some installation components (such as labour). Consequently, mature technologies, which have limited prospect of installation cost reductions, are the most impacted by this new escalation factor (e.g., gas and coal technologies).
4.2 Global generation mix
The rate of global technology deployment is the key driver for the rate of reduction in technology costs for all non-mature technologies. However, the generation mix is determined by technology costs. Recognising this, the projection modelling approach simultaneously determines the global generation mix and the capital costs. The projected generation mix consistent with the capital cost projections described in the next section is shown in Figure 4-1.
8 It is referred to as an easement cost index in that document.
GenCost 2025-26 | 31
Figure 4-1 Projected global electricity generation mix in 2030 and 2050 by scenario
The technology categories displayed are more aggregated than in the model to improve clarity. Solar includes solar thermal and solar photovoltaics.
Current policies and Global NZE post 2050 have the lowest electrification because they are a slower decarbonisation pathway. However, they have the least energy efficiency and industry transformation which supports electricity demand growth9. Global NZE by 2050 has high vehicle electrification and high electrification of other industries including hydrogen. However, it also has high energy efficiency and industry transformation which partially offsets these sources of new electricity demand growth. Current policies and Global NZE post 2050 have similar demand in both 2030 and 2050 indicating these opposing drivers are somewhat in balance in those scenarios over time with a slight dip in 2050 for Global NZE post 2050 due to stronger energy efficiency improvements. However, Global NZE by 2050 has the highest electricity demand in both 2030 and 2050 indicating electrification is the stronger driver throughout the projection period.
Figure 4-2 shows the contribution of each hydrogen production technology in each scenario indicating the Global NZE by 2050 scenario is assumed to experience a significant growth in electrolysis hydrogen production and modest growth in the other two scenarios. Note that the IEA’s 2025 estimates of hydrogen demand which are used to guide GenCost global assumptions, have decreased relative to previous outlooks across all scenarios, but most notably in the Global NZE post 2050 scenario.
9 Economies can reduce their emissions by reducing the activity of emission intensive sectors and increasing the activity of low emission sectors. This is not the same as improving the energy efficiency of an emissions intensive sector. Industry transformation can also be driven by changes in consumer preferences away from emissions intensive products.
0100002000030000400005000060000700008000090000Current policiesGlobal NZE post2050Global NZE by2050Current policiesGlobal NZE post2050Global NZE by205020302050Generation (
TWh)BECCSCoalCoal CCSGasGas CCSHydroNuclearOilSolarWind onshoreWind offshoreOther renewables
32 | CSIRO Australia’s National Science Agency
Figure 4-2 Global hydrogen production by technology and scenario, Mt
Current policies and Global NZE post 2050 have the lowest non-hydro renewable share but still relatively high at 58 and 69% of generation respectively by 2050 compared to 81% in Global NZE by 2050. Growth in hydro generation is constrained by land use competition and coal by emissions constraints. Consequently, gas with and without CCS and nuclear make up most of the remaining new generation by 2050. Gas with CCS is preferred to coal with CCS given the relatively lower capital cost and lower emissions intensity of that technology. Nuclear is around 9% of generation in 2030 and is steady or, in the worst case, declining to 6.7% generation share by 2050 reflecting its relatively slower development and installation rate while electrification causes global demand to grow rapidly in the next three decades.
The Global NZE by 2050 scenario is close to but not completely zero emissions by 2050. All generation from fossil fuel sources is with CCS accounting for 2% of generation by 2050. Offshore wind features strongly in this scenario at 11% of generation by 2050. Renewables other than hydro, biomass, wind and solar are 5% of generation in 2050. The greater deployment of renewables and CCS in this scenario leads to lower renewable and CCS costs. CCS costs are also impacted by the use of CCS in hydrogen production and other industries.
4.3 Changes in capital cost projections
This section discusses the changes in cost projections to 2055 compared to the 2024-25 projections. For mature technologies, differences mainly reflect any changes in current costs, an assumed return to normal costs by 2035 and land and construction costs increases thereafter.
050100150200250300350400Current policiesGlobal NZE post2050Global NZE by2050Current policiesGlobal NZE post2050Global NZE by205020302050Hydrogen production (
Mt)ElectrolysisSteam methane reformingSteam methane reforming with CCS
GenCost 2025-26 | 33
Less mature technologies include learning components in addition to the land and construction cost escalators. For technologies with high learning potential, the cost reduction from learning more than offsets the escalation factors for most of the projection period. For those with lower learning potential, the cost changes may cancel one another out or the escalation factors may dominate the trend.
Data tables for the full range of technology projections are provided in Appendix B and can be downloaded from CSIRO’s Data Access Portal10.
4.3.1 Black coal ultra-supercritical
The updated cost of black coal ultra-supercritical plant in 2025 has been sourced from GHD (2026). This included a substantial increase based on updates to standard software used to model coal generation capital costs. From 2025, the capital cost is assumed to be constant in real terms in 2026 (increasing in nominal terms) and return to levels consistent with ultra-supercritical prior to the COVID-19 pandemic by 2035 adjusted for changes in construction costs. Black coal ultra-supercritical is treated in the projections as a learning technology. However, global new building of ultra-supercritical coal is limited due to climate change policies and consequently the cost reduction achieved from deployment via the learning rate is low.
Figure 4-3 Projected capital costs for black coal ultra-supercritical by scenario compared to 2024-25 projections
10 Search GenCost at https://data.csiro.au/collections
0100020003000400050006000700080002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
34 | CSIRO Australia’s National Science Agency
The outlook for costs in all scenarios is increasing due to increasing land and installation costs. Installation costs are rising faster the stronger the climate policy ambition of the scenario reflecting a stronger rate of electricity sector construction activity.
4.3.2 Coal with CCS
The capital cost of black coal with CCS from 2025 to 2035 has been updated according to the approach outlined in the beginning of this section. Thereafter, the capital cost of the mature parts of the plant reflects assumed land and installation cost increases. For the CCS components, in addition to these changes in land and installation costs, changes in equipment costs are a function of global deployment of gas and coal with CCS, steam methane reforming with CCS and other industry applications of CCS. Compared to the 2024-25 projections, global CCS deployment has not significantly changed.
Current policies has no uptake of steam methane reforming with CCS in hydrogen production. Consequently, any equipment cost reductions from the late 2030s are mainly driven by the deployment of CCS in other industries. While black coal with CCS benefits from co-learning from deployment of CCS in non-electricity industries, there is only a negligible amount of generation from black coal with CCS throughout the projection period.
Figure 4-4 Projected capital costs for black coal with CCS by scenario compared to 2024-25 projections
Global NZE by 2050 and Global NZE post 2050 take up CCS in hydrogen production and both gas and coal electricity generation (although gas generation with CCS is significantly more preferred and the amount is much lower in Global NZE post 2050). The total CCS deployment in electricity generation and hydrogen production is higher in Global NZE by 2050. However, CCS deployment in other industries is higher in Global NZE post 2050. Subsequently, those scenarios experience a similar amount of equipment cost reduction by 2050 but with stronger local construction cost
020004000600080001000012000140002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 35
increases in Global NZE by 2050. In Current policies, equipment cost reductions are not significant, but installation cost increases are lower than the Global NZE scenarios.
A first of a kind premium, in addition to the costs shown, will likely apply when coal with CCS is deployed in Australia for the first time.
4.3.3 Gas combined cycle
GHD (2026) have included negligible real change in gas combined cycle costs for 2025 which is a change compared to real increases the previous two years. In the next two years costs are expected to increase due to the impacts of data centre demand on gas turbine prices and then slowly return to normal by 2035. After the return to normal period, because gas combined cycle is classed as a mature technology for projection purposes, its change in capital cost is governed only by assumed increases in land and installation costs for all scenarios. Consistent with the need for greater construction activity the stronger the climate policy ambition, combined cycle gas costs are highest in Global NZE by 2050.
Figure 4-5 Projected capital costs for gas combined cycle by scenario compared to 2024-25 projections
4.3.4 Gas with CCS
The current cost for gas with CCS has been revised upwards for the 2025-26 projections and is assumed to increase further in the next two years due to the impact of data centre demand on gas turbine prices. From 2028, costs decline to 2035 based on our return to normal assumptions during this period. The relativities between the scenarios reflect the changes in land and installation costs increases and differences in global deployment in electricity generation, hydrogen production and other industry uses of CCS. Global NZE by 2050 has the highest total
0500100015002000250030002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
36 | CSIRO Australia’s National Science Agency
deployment of all CCS technologies. Subsequently, the equipment component of gas with CCS is lower by 2050 in that scenario and this results in total costs being lower from the late 2030s. CCS equipment costs are highest cost where CCS deployment is lowest in Current policies and Global NZE post 2050. While these two scenarios are assumed to experience lower construction costs increases, they remain highest cost overall by 2050.
The IEA CCS database11 indicates there are over 100 planned electricity related projects which are yet to make a financial investment decision and around 8 under construction. One is operational. Given the current state of the pipeline of projects, significant global deployment of CCS is not expected until after 2030.
A first of a kind premium, in addition to the costs shown, will likely apply when gas with CCS is deployed in Australia for the first time.
Figure 4-6 Projected capital costs for gas with CCS by scenario compared to 2024-25 projections
4.3.5 Gas open cycle (small and large)
Figure 4-7 shows the 2025-26 cost projections for small and large open cycle gas turbines. All new gas turbine projects are expected to include the capability for hydrogen blending and eventual conversion to hydrogen firing when hydrogen supply becomes more readily available and lower cost. This is in addition to the existing ability to use liquid fuels such as diesel or renewable diesel. However, it is possible that some plants will only ever use natural gas during their life. It depends on the market conditions and climate policy during their operation. The small open cycle gas
11 CCUS Projects Database - Data product - IEA, including only those projects that include storage
0100020003000400050006000700080002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 37
technology is designed with a maximum 35% hydrogen blend. The large size is designed for 10%. However, GHD (2026) also provides costs for higher and lower blends at both sizes. This assumption of hydrogen readiness adds a negligible premium to gas open cycle capital costs.
The GHD (2026) report provides additional details for the unit sizes and total plant capacity that defines the small and large sizes. Gas open cycle costs are increasing rapidly due to strong growth in data centre demand outpacing manufacturing supply capability. Costs are expected to increase further in the next two years as manufacturing capacity remains below demand. Costs are not projected to return to normal until 2035. For the remainder of the projection period, there are no improvements in equipment costs because of the maturity of the technology, and so the assumed land and installation cost increases result in a rising trend in costs post-2035. Capital costs are highest under the Global NZE scenarios reflecting the higher installation costs associated with the greater construction activity of those scenarios
Figure 4-7 Projected capital costs for gas open cycle (small) by scenario compared to 2024-25 projections
4.3.6 Nuclear SMR
For the next five years, costs are based on the planned Darlington SMR project in Canada which consists of four 300MW units for a total cost of C$20.9b. Costs are expected to be highest for the first unit but lower for each subsequent unit and this is captured in the cost trajectory. Unlike large-scale nuclear, to convert Darlington nuclear SMR costs to Australian dollars the method only included an exchange rate conversion. That is, no allowance has been made for differences in construction costs between Canada and Australia. The difference in approach is justified based on the high level of commercial immaturity of nuclear SMR outweighing any other uncertainties in the cost estimate. The Darlington project cost conversion also includes removal of interest during construction in the capital cost compared with the 2024-25 projections.
05001000150020002500300035002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistorySmallLarge
38 | CSIRO Australia’s National Science Agency
The rate of cost reductions after the Darlington project is calculated as function of deployment of other global nuclear SMR projects, to a greater or lesser degree depending on the global scenario and some known projects. Capital costs only improve in the 2040s for the Current policies scenario due to a lack of additional deployment of projects in the 2030s. The Global NZE scenarios achieve a greater level of deployment of nuclear SMR in the 2030s owing to a stronger commitment to addressing climate change.
Nuclear SMR equipment cost reductions may be partly driven by modular manufacturing processes. Modular plants reduce the number of unique inputs that need to be manufactured. Assumed increases in land and installation costs are responsible for increases in Australian nuclear SMR costs in the 2040s and 2050s.
A first of a kind premium, in addition to the costs shown, will likely apply when nuclear SMR is deployed in Australia for the first time.
Figure 4-8 Projected capital costs for nuclear SMR by scenario compared to 2024-25 projections
4.3.7 Large-scale nuclear
Given Australia has no experience building large scale nuclear, we base costs on South Korean nuclear building costs adjusted for the relative costs of building ultra-supercritical coal in each country. Given the cost of building ultra-supercritical coal in Australia has increased, nuclear costs are also revised upwards. From a more direct perspective, given all steam turbine costs have gone up, then nuclear costs are also impacted. However, like other technologies, large-scale nuclear capital costs are assumed to return to their underlying costs by 2035 after constant real costs in 2026 (rising in nominal terms).
050001000015000200002500030000350002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 39
Large-scale nuclear is treated as a mature technology and therefore is not assigned any learning rate whereby cost reductions are achieved as a function of deployment. Instead, large-scale nuclear costs increase after 2035 due to the assumed increases in land and installation costs that impact all technologies.
There is some uncertainty in the literature about whether large-scale nuclear is a learning technology or not. There are many new designs for nuclear generation and so it is not a settled technology in the way we might consider steam turbines. Even settled technologies still incrementally change. However, our reluctance to assign a learning rate to large-scale nuclear reflects two issues. First, an assigned learning rate would have little impact because it is difficult for any mature technology to double its global capacity which is the required trigger to receive the benefit of an assigned learning rate (see Appendix A for an explanation of the learning rate function). Second, new designs for large-scale nuclear have not always delivered cost reductions. Therefore, our projection reflects a nuclear industry that mostly consolidates construction around proven designs.
A first of a kind premium, in addition to the costs shown, will likely apply when large-scale nuclear is deployed in Australia for the first time.
Figure 4-9 Projected capital costs for large-scale nuclear by scenario compared to 2024-25 projections
4.3.8 Solar thermal
The starting cost for solar thermal has decreased in GHD (2026) and this results in lower costs by 2055 relative to 2024-25 in some scenarios. While greater deployment and equipment cost reduction is generally aligned to the stronger abatement scenarios, local construction costs increase with stronger abatement. These higher construction costs are the stronger factor over
0200040006000800010000120002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
40 | CSIRO Australia’s National Science Agency
time for this technology. Consequently, long term costs are higher under the Global NZE scenarios and lowest under Current policies.
Solar thermal systems consist of the combination of solar mirror field, thermal storage and power blocks that are sized in varying ratios according to the location and market signals that prevail. Each such configuration will have a different capital cost. As a consequence, the baseline configuration represented in the capital cost projection data is not the same as the configurations used to calculate the LCOEs in Section 5.
A first of a kind premium, in addition to the costs shown, will likely apply when solar thermal is deployed in Australia for the first time.
Figure 4-10 Projected capital costs for solar thermal with 14 hours storage compared to 2024-25 projections
Historical data includes costs for some 15 and 16 hour project designs
4.3.9 Large-scale solar PV
Large-scale solar PV costs have been revised upwards for 2025 based on GHD (2026). This represents a reversal of the last two historical years of cost reductions but likely reflects cost volatility rather than a new trend. As a result of past cost reductions for this technology, unlike other technologies, we do not impose any additional cost reduction related to recovery from the global inflationary pressures. All cost reductions in the projection are due to learning through deployment. Weaker abatement policies and lower electricity generation in Current policies and Global Net Zero Post 2050 result in less solar PV deployment and less cost reduction. Global Net Zero by 2050 has the largest deployment with costs declining to around $600/kW.
The final minimum cost level for solar PV is difficult to predict because, unlike other technologies, and notwithstanding recent inflationary pressures, the historical learning rate for solar PV has not significantly slowed. The modular nature of solar PV appears to be the main point of difference in
0100020003000400050006000700080009000100002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 41
explaining this characteristic. High global manufacturing capacity that has kept up with or exceeded demand has also been a significant feature of this technology.
All scenarios include increases in installation costs in Australia and this also contributes to the narrowing of the differences between the scenarios over time. Installation costs are assumed to grow faster the stronger the global climate policy ambition due to stronger construction activity.
Figure 4-11 Projected capital costs for large-scale solar PV by scenario compared to 2024-25 projections
4.3.10 Rooftop solar PV
The current costs for rooftop solar PV systems are lower than was projected for 2025 in the 2024-25 GenCost report. Rooftop solar PV is sold across a broad range of prices12 and consequently this data is best interpreted as a mean and may not align with the lowest cost systems available. The cost is before available subsidies and on the basis of the direct current power rating of the system whereas large-scale solar PV and all other generation technologies are on an alternating current power rating basis.
Rooftop solar PV benefits from co-learning with the components in common with large scale PV generation and is also impacted by the same drivers for variable renewable generation deployment across scenarios. However, the rate of capital cost reduction in each scenario is slower than large-scale solar PV because we have assumed a low learning rate on the installation or local learning component for rooftop solar. This reflects that Australia already has a very high degree of experience in installing rooftop solar so there are less opportunities to reduce the cost
12 The Cost of Solar Panels - Solar Panel Price | Solar Choice
050010001500200025002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
42 | CSIRO Australia’s National Science Agency
of installation compared to large-scale solar PV. The Global NZE by 2050 scenario has a higher learning rate on installation than the other scenarios and this allows for a stronger cost reduction trend.
Rooftop solar PV installation costs are also impacted by the general increase in installation costs that apply to all technologies.
Figure 4-12 Projected capital costs for rooftop solar PV by scenario compared to 2024-25 projections
4.3.11 Onshore wind
As the historical data indicates onshore wind is one of the technologies which has been most impacted by recent global inflationary pressures. The updated GHD (2026) data indicates that the costs pressures are stabilising. However, due to the Iran war, no further improvement in costs is expected in 2026.
To recognise the more difficult circumstances for the onshore wind industry locally and globally, our assumption is that capital costs of onshore wind will not return to its normal cost path until 2035 in all scenarios. After 2035, wind costs are projected to decline only a modest amount. Global equipment cost reductions from learning are offset by local increases in land and installation costs. While equipment costs fall the most in stronger climate policy ambition scenarios, these scenarios also experience the strongest increase in installation costs due to greater construction activity. Consequently, these global and local changes in costs tend to offset one another resulting in little difference between the three scenarios by 2055.
050010001500200025002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 43
Figure 4-13 Projected capital costs for onshore wind by scenario compared to 2024-25 projections
4.3.12 Fixed and floating offshore wind
Fixed and floating offshore wind are represented separately in the projections. Our general approach is not to include similar technologies because of model size limits and because the model will usually choose only one of two similar technologies to deploy, therefore adding no new insights. However, while the two offshore technologies have a lot of common technology, floating wind is less constrained in terms of the locations in which it can be deployed. As the global effort to reduce greenhouse gas emissions looks increasingly to electricity as an energy source, many countries will be seeking to use technologies that have fewer onshore siting conflicts. Fixed offshore wind is the lowest cost offshore technology, but its maximum deployment is limited by access to seas of a maximum depth of around 50-60 metres13 and any navigation, marine conservation or aesthetic issues within those zones. Floating offshore wind can be deployed at much greater depths increasing its potential global deployment and providing a unique reason to select the technology.
Figure 4-14 presents projections for both fixed and floating compared to the 2024-25 projection. The current costs for both types of offshore wind are provided in GHD (2026). The updated current capital costs are lower than projected in 2024-25 for floating offshore wind and higher than projected for fixed offshore wind. Post 2025, offshore wind capital costs are not adjusted for inflationary pressures in the same way as other technologies because fixed offshore wind has already recovered based on the average global data which informs the historical series. However,
13 This is more an economic than absolute technical limit.
050010001500200025003000350040002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
44 | CSIRO Australia’s National Science Agency
it is likely that technology prices are higher for some regions and manufacturers. Australia is not likely to deploy offshore wind before 2030 and therefore GenCost will continue to be required to rely on global sources of offshore wind cost data until then. A first of a kind premium, in addition to the costs shown, will likely apply when offshore wind is deployed in Australia for the first time.
Figure 4-14 Projected capital costs for fixed and floating offshore wind by scenario compared to 2024-25 projections
Fixed offshore wind costs are higher than projected in 2024-25 but this mostly reflects the higher current cost rather than a change in global deployment. Floating offshore wind projections have a greater range of cost reduction pathways, and this reflects their relative immaturity and greater potential global deployment.
Offshore wind is not as impacted as other technologies on land costs but does require some onshore land to connect to the grid. Offshore wind costs are impacted by the new assumptions with regards to increasing installation costs.
4.3.13 Battery storage
Current 2025 costs of battery storage fell in line with the fastest cost reduction projected in 2024-25 and the updated cost projections continue to allow for a modest range of cost reductions after 2025 but slower than the most recent year. The costs shown in Figure 4-15 are for a 2-hour duration battery (total battery cost including battery and balance of plant). Given the 2025 cost reduction takes batteries back to below their pre-pandemic levels we do not impose any additional reduction beyond the learning projected by the modelling.
0100020003000400050006000700080009000100002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 2050 (fixed)2024-25 Global NZE by 2050 (fixed)2025-26 Global NZE post 2050 (fixed)2024-25 Global NZE post 2050 (fixed)2025-26 Current policies (fixed)2024-25 Current policies (fixed)2025-26 Global NZE by 2050 (floating)2024-25 Global NZE by 2050 (floating)2025-26 Global NZE post 2050 (floating)2024-25 Global NZE post 2050 (floating)2025-26 Current policies (floating)2024-25 Current policies (floating)History
GenCost 2025-26 | 45
Figure 4-15 Projected total capital costs for 2-hour duration batteries by scenario (battery and balance of plant)
The projections use different learning rates by scenario to reflect the uncertainty as to whether they will be able to continue to achieve their high historical cost reduction rates (notwithstanding the pandemic period). Historical cost reductions have mainly been achieved through deployment in industries other than electricity such as in consumer electronics and electric vehicles. Global electric vehicle uptake has been updated with inputs from IEA (2025). While these other uses are important, small- and large-scale stationary electricity system applications are growing globally. Under the three global scenarios, batteries have a large future role to play in supporting variable renewables alongside other storage and flexible generation options and in growing electric vehicle deployment.
Battery deployment is strongest in the Global NZE by 2050 scenario reflecting stronger deployment of variable renewables, which increases electricity sector storage requirements. Together with an assumed high learning rate this leads to the fastest cost reduction. The remaining scenarios have more moderate cost reductions reflecting a reduced requirement for stationary storage and assumed lower learning rates. All projections are impacted by assumed increases in installation costs. However, for batteries, the learning effects more than offset this factor leading to declining cost trajectories.
A breakdown of battery pack and balance of plant costs for various storage durations are provided in Appendix B.
GHD (2026) has included current costs for small-scale batteries, designed to be installed in homes. They are estimated at $11,000 for a 5kW/10kWh system or $1100/kWh, including installation but excluding subsidies. This is around twice the cost of large-scale battery projects per kWh.
0200400600800100012002015202020252030203520402045205020552025 $/kWh2024-25 Current policies2024-25 Global NZE post 20502024-25 Global NZE by 20502025-26 Current policies2025-26 Global NZE post 20502025-26 Global NZE by 2050History
46 | CSIRO Australia’s National Science Agency
However, larger household batteries are achieving lower per kWh costs more consistent with large-scale costs.
4.3.14 Pumped hydro energy storage
GHD (2026) has provided a reassessment of pumped hydro energy storage (PHES) costs and this is the main driver of differences in the cost outlook compared to 2024-25. See GHD (2026) for more discussion. PHES is a mature technology and receives the same increase in installation costs as other technologies which is the main driver for the increasing cost trend post 2030. Unlike the other technologies, all three scenarios assume costs return to normal by 2030 (rather than in 2035). This reflects the already large downgrade in costs in 2025. Site variability is also a great source of variation in PHES costs and is separately addressed by GHD (2026) and AEMO external to GenCost.
The cost trajectory shown in Figure 4-16 is for a 24-hour duration storage design. Costs for 10-hour, 48-hour and 160-hour durations are also included in this report (Appendix B).
Figure 4-16 Projected capital costs for pumped hydro energy storage (24-hour) by scenario
4.3.15 Other generation technologies
There are several technologies that are not commonly deployed in Australia but may be important from a global energy resources perspective or as emerging technologies. These additional technologies are included in the projections for completeness and discussed below. They are each influenced by revisions to current costs which have generally experienced an increase in capital costs for 2025 with the exception of fuel cells. Reflecting the infrequency with which these technologies are built, the increases for some technologies mostly represent theoretical increases
0100020003000400050006000700080002015202020252030203520402045205020552025 $/kW2025-26 Global NZE by 20502025-26 Global NZE post 20502025-26 Current policies2024-25 Global NZE by 20502024-25 Global NZE post 20502024-25 Current policiesHistory
GenCost 2025-26 | 47
in costs if they had been built based on the general increase in infrastructure building costs. The flat trend in 2026 followed by a downward trend to 2035 has been included using the same methodology for the technologies above. The projections also include increasing land and installation costs for biomass with CCS and fuel cells (wave and tidal/ocean current are excluded due to insufficient data).
Figure 4-17 Projected technology capital costs under the Current policies scenario compared to 2024-25 projections
Current policies
Biomass with CCS is deployed at a negligible level in the Current policies scenario because the climate policy ambition is not strong enough to incentivise significant deployment. Cost changes after 2035 reflect co-learning from other CCS technologies which are deployed in electricity generation and in other sectors but these are more than offset by increasing installation costs. There is also no significant deployment of tidal or wave technology reflecting the lack of climate policy ambition. Fuel cells are lower owing to updated current costs and minor deployment in the late 2040s.
Global NZE by 2050
Biomass with CCS has the highest deployment in Global NZE by 2050 scenario but this scenario also has the strongest increase in installation costs. Biomass with CCS is an important technology in some global climate abatement scenarios if the electricity sector is required to produce negative abatement for other sectors. However, we are not able to model that scenario with GALLME. GALLME only considers the value of this technology to the electricity sector’s own abatement task and from that perspective alone, biomass with CCS is a relatively high-cost generation technology.
050001000015000200002500030000202020252030203520402045205020552025 $/kWTidal/Ocean currentFuel cellWaveBiomass with CCS2024-25 Tidal/Ocean current2024-25 Fuel cell2024-25 Wave2024-25 Biomass with CCS
48 | CSIRO Australia’s National Science Agency
Wave energy is deployed at a modest level in the 2040s leading to some costs reductions during those periods. Fuel cells are lower owing to updated current costs.
Figure 4-18 Projected technology capital costs under the Global NZE by 2050 scenario compared to 2024-25 projections
Global NZE post 2050
Biomass with CCS, wave and tidal energy are not deployed at significant scale in this scenario. However, fuel cell technologies are favoured in this scenario with modest but steady deployment through to 2055. This result is similar to the Current policies scenarios but with a stronger reduction in fuel cell costs. This reflects the weaker abatement policy leading to stronger reliance on more mature abatement technologies.
050001000015000200002500030000202020252030203520402045205020552025 $/kWTidal/Ocean currentFuel cellWaveBiomass with CCS2024-25 Tidal/Ocean current2024-25 Fuel cell2024-25 Wave2024-25 Biomass with CCS
GenCost 2025-26 | 49
Figure 4-19 Projected technology capital costs under the Global NZE post 2050 scenario compared to 2024-25 projections
4.3.16 Hydrogen electrolysers
Hydrogen electrolyser costs have decreased in 2025 for both proton-exchange membrane (PEM) and alkaline electrolysers based on GHD (2026). Alkaline electrolysers remain lower cost than PEM electrolysers but their costs are becoming closer together.
The key advantage of PEM electrolysers is their wider operating range which gives them a potential advantage in matching their production to low-cost variable renewable energy generation. If the costs of electrolysers fall sufficiently, capital costs would become less significant in total costs of hydrogen production. This development could make it attractive to sacrifice some electrolyser capacity utilisation for lower energy costs (by reducing the need to deploy storage in order to keep up a minimum supply of generation). Under these circumstances, the more flexible PEM electrolysers could be preferred if their costs are low enough.
Deployment of electrolysers and subsequent cost reductions are projected to be greatest in the Global NZE by 2050 scenario with a lesser and similar amount of change expected in Current policies and Global NZE post 2050. However, given stronger assumed increases in installation costs, Global NZE post 2050 is the higher cost of the two scenarios.
The 2025-26 electrolyser cost projections are significantly higher than in 2024-25. This reflects the significantly reduced outlook for hydrogen production from electrolysers that has occurred due to updating the global hydrogen demand projections from IEA (2025). Lower deployment means lower cost reductions under the learning based projection approach used in GALLM (see appendix A).
050001000015000200002500030000202020252030203520402045205020552025 $/kWTidal/Ocean currentFuel cellWaveBiomass with CCS2024-25 Tidal/Ocean current2024-25 Fuel cell2024-25 Wave2024-25 Biomass with CCS
50 | CSIRO Australia’s National Science Agency
Figure 4-20 Projected technology capital costs for alkaline and PEM electrolysers by scenario, compared to 2024-25
0500100015002000250030003500400045005000202020252030203520402045205020552025 $/kW2025-26 Current policies Alkaline2025-26 Current policies PEM2025-26 Global NZE by 2050 Alkaline2025-26 Global NZE by 2050 PEM2025-26 Global NZE post 2050 Alkaline2025-26 Global NZE post 2050 PEM2024-25 Current policies Alkaline2024-25 Current policies PEM2024-25 Global NZE by 2050 Alkaline2024-25 Global NZE by 2050 PEM2024-25 Global NZE post 2050 Alkaline2024-25 Global NZE post 2050 PEM
GenCost 2025-26 | 51
5 Levelised cost of electricity analysis
5.1 LCOE definition
Levelised cost of electricity (LCOE) data is an electricity generation technology comparison metric. It is the total unit costs a generator must recover to meet all its costs including a return on investment14. Modelling studies such as AEMO’s Integrated System Plan (AEMO, 2024) do not require or use LCOE data15. LCOE is a simple screening tool for quickly determining the relative competitiveness of electricity generation technologies. It is not a substitute for detailed project cashflow analysis or electricity system modelling which both provide more realistic representations of electricity generation project operational costs and performance.
The standard method of calculating LCOE does not include all of the additional costs required to deliver reliable electricity supply from variable renewable energy (VRE) generation, particularly when the share of VRE generation is high. The two key VRE technologies are wind and solar photovoltaics (PV). To address this issue of additional VRE costs, since 2019, GenCost has deployed a two-step methodology which separately calculates the VRE integration costs in an electricity system model and then adds them back into the standard LCOE for wind and solar photovoltaics. This method was devised after a thorough literature review in Graham (2018) but has received an update in this 2025-26 report.
5.2 Change in method for estimating the cost of reliable high VRE share generation
This 2025-26 GenCost report revises the methodology for estimating the cost of high VRE electricity systems in response to stakeholder feedback. Through various submissions over several years, stakeholders requested that the methodology be revised in two ways:
1.
Use a System Levelised Cost of Electricity (SLCOE) approach. A SLCOE takes the cost of electricity directly from an electricity system model by dividing the system costs associated with existing and new technology deployments by the total useful electricity supply in a given year. This concept is also equivalent to the average annual unit cost of electricity for a given electricity system boundary16. Calculation of SLCOE is more direct than the previous approach which required setting up a baseline and identifying additional integration costs by calculating differences in costs for alternative levels of VRE generation share. Another advantage is that SLCOE also provides a system perspective which inherently requires a
14For a description the LCOE formula and the application of the formula go to CSIRO’s Data Access Portal and download the latest Excel file that accompanies this report. CSIRO Data Access Portal
15 LCOE is a measure of the long run marginal cost of generation which could partly inform generator bidding behaviour in a model of the electricity dispatch system. However, in such cases, it would be expected that the LCOE calculation would be internal to the modelling framework to ensure consistency with other model inputs rather than drawn from separate source material.
16 In this case we are concerned with generation and transmission
52 | CSIRO Australia’s National Science Agency
bundle of technologies. This is a more useful perspective than comparing single technologies against one another using LCOE since very few systems use a single technology (the exception being remote power systems but even these are increasingly combining multiple technologies.)
2.
Provide greater transparency with regard to the data inputs and the modelling system used.
Separate from these two key items of stakeholder feedback, it was also observed over several years that while public interest in the future generation mix remains high, and the GenCost project’s primary goal is to support electricity system modelling by providing cost data, the amount of published electricity system modelling had not increased. It is hypothesised that the high complexity and cost of creating and applying commercial electricity system modelling tools may have been a significant contributor to this outcome. With this context, and to address all three issues simultaneously, a new open source electricity system model was created to calculate SLCOE, replacing the previous method for estimating the integration costs for solar PV and wind.
The new electricity system model is a simplified version of the larger commercial models used by CSIRO and other organisations. With the needs of other potential users in mind, the model was designed to be as short and fast solving as possible whilst not significantly reducing accuracy regarding the estimation of electricity costs and the generation mix. The final model design and the simplifications considered in the development phase are described in Graham et al. (2025). The model is of the National Electricity Market (NEM), excluding Western Australia and Northern Territory. The exclusion of these regions is a simplification in itself but also reflects the fact that these jurisdictions do not provide enough publicly available information about their electricity systems to easily construct an open source model.
Additional model details, software code and all input data can be downloaded at https://data.csiro.au/collection/csiro:71289. The model has been named Simple Electricity Model or SEM. Despite the goal of simplification, the model still requires access to and knowledge of specialised software and solvers and is targeted at users with a university level of knowledge who will likely have most success in following the equations and code. This is because universities typically provide education on linear programming across a wide variety of degree courses as well as access to free software and solver licences. Simpler spreadsheet type models or calculations were considered as an alternative but ultimately are not accurate enough to be relied on to estimate system costs due to their inability to consider and optimise all of the available system configurations.
5.3 The 2030 SLCOE, existing capacity and electricity futures prices
The main feedback that the new SLCOE method received on the consultation draft of this report was that some stakeholders are uncomfortable with or uncertain about how existing capacity should be treated in an SLCOE calculation that is focussed on a near term year such as 2030. Applying SLCOE to 2050 is less complicated because most existing capacity will be retired and to some extent can be ignored (and this is the approach taken in academic articles such as Idel (2022)). However, in a 2030 SLCOE existing capacity cannot be ignored. The model must recognise the availability of existing capacity if it is to realistically represent the 2030 electricity system.
GenCost 2025-26 | 53
Views differ on how much of the historical cost of each item of existing capacity should contribute to system costs. Aside from transmission which has a specified regulatory system, there is no public record of how much of the original investment plus reasonable return has already been received by the owner of generation and storage assets before 2030. As such, any attempt to include the cost of existing capacity will be based on low quality data. Stakeholders offered suggestions for costs of existing projects that should be included in the system cost but no general consistent approach for how to cost existing capacity.
As the GenCost report is intended to be updated each year including 2030, this implementation issue in regard to SLCOE can only grow larger as existing capacity will become an increasing share of total capacity. This has led to the conclusion that with the practical limitations of data, SLCOE is not appropriate for 2030 or short term system costing in general.
There are alterative modelling approaches that could project the cost of electricity in 2030 by forecasting electricity spot market prices. However, these are large, commercial models and consequently they would not meet the goals of simplicity and transparency. An alternative approach that does meet these goals is to look to electricity futures prices. These are a hedging product for market participants and as such they provide reasonable indicator of the market participants’ expectation of electricity prices. Electricity futures prices are available for around 3-4 years ahead from https://www.asxenergy.com.au/. The baseload contract includes all hours of the year and so is equivalent in scope to an SLCOE cost for a specific year. Electricity futures contracts also include a daily evening peak product. This will be useful as a comparison point for peaking technologies included in GenCost whereas SLCOE results were only relevant as a comparison point for bulk energy technologies.
5.4 2050 SLCOE scenarios
The 2050 LCOE focusses on an emissions intensity target range in 2050 that supports achieving the policy of net zero emissions by 2050. This is the most common type of 2050 policy in place at the national, state and corporate level. However, stakeholders interested in other scenarios can explore them with the open source model provided.
The 2050 SLCOE considers either mature firmed renewables only or mature firm renewables plus any of three groups of technologies: floating and fixed offshore wind, coal and gas with carbon capture and storage and large-scale and small modular reactor nuclear. Each additional technology group is modelled separately because they each are assumed to require two first-of-a-kind (FOAK) projects to be built before they can access the standard technology costs. As such each technology group represents a unique scenario creating three renewable plus scenarios. It would not be advisable to design a scenario where all technologies are built as it would be prohibitively expensive to build FOAK projects for all technologies. After building the two FOAK projects in each of the technology group scenarios the model determines whether to build any more of that technology at the standard cost based on whether additional capacity of that technology contributes to achieving a least cost electricity system.
The 2050 scenarios must meet emission intensity targets of 0.0025tCO2e/MWh up to 0.20tCO2e/MWh. On their own these numbers mean little and so Table 5-1 provides more context and scenario names. It is appropriate to explore a range of electricity sector emissions intensity
54 | CSIRO Australia’s National Science Agency
targets because there is no strict policy on the degree to which any sector must decarbonise in a net zero emissions policy world. To minimise greenhouse gas emissions abatement costs, ideally emission reduction takes place in each sector up to the point where no less expensive abatement may be found in any other sectors, including offset sectors such as land use, land use change and forestry. The value of 0.02tCO2e/MWh is most closely aligned with AEMO’s Step Change scenario in the 2024 and draft 2026 Integrated System Plan (ISP) modelling results, after adjusting for the ISP only reporting direct emissions and GenCost modelling being on a full fuel cycle basis (that is, inclusive of fugitive emissions in fuel supply). Other emissions intensity scenarios serve as minimum and maximum costs, helping to define the range of possible electricity systems costs in 2050.
Table 5-1 2050 full fuel cycle greenhouse gas emissions intensity target scenarios and their meaning 2050 emissions intensity target scenario Scenario name Meaning 0.20tCO2e/MWh NoProgressToNetZero Consistent with achieving the 2030 82% renewables target policy target and holding the emissions intensity constant to 2050. This is not consistent with achieving net zero by 2050 but serves as a reference case cost against which reducing the emissions intensity beyond 2030 can be measured.
0.10tCO2e/MWh
WeakNetZero
A halving of the 2030 emissions intensity by 2050. Only weakly consistent with the Australia’s net zero by 2050 policy, requiring significantly more abatement in non-electricity sectors. 0.05tCO2e/MWh ModerateNetZero A 75% reduction in the 2030 emissions intensity by 2050. In the range of what is required for the electricity sector to meet Australia’s net zero by 2050 policy
0.02tCO2e/MWh
StrongNetZero
The emissions intensity consistent with AEMO’s 2024 and draft 2026 Integrated System Plan Step Change scenario. Likely to be what is required to meet Australia’s net zero by 2050 policy 0.0025tCO2e/MWh VeryStrongNetZero Almost eliminating emissions from the electricity sector (just under a million tonnes remain). This likely exceeds what is required to meet Australia’s net zero by 2050 policy but serves to define the maximum cost of electricity associated with emissions abatement.
GenCost 2025-26 | 55
A zero emission scenario is not included because, on a full fuel cycle basis only the mature firm renewables technology mix on its own or plus-offshore wind can achieve this outcome. The plus-nuclear and plus-CCS options have upstream emissions in their fuel supply chain and plus-CCS downstream in the CO2 released that is not captured.
AEMO currently makes available 15 historical weather years of demand and renewable production data. It is possible to simulate all of these. However, for the sake of brevity and because for reliability purposes sufficient renewables and other supporting technologies must be deployed to deal with the worst weather conditions, all scenarios only use the single most costly weather year which is 2011. Graham et al. (2025) provides more detail on the distribution of costs across weather years.
5.5 SLCOE estimates
5.5.1 Data alignment
The results below are based on the capital cost projections of generation and storage technologies in this final GenCost report and on transmission costs and other electricity sector inputs available in AEMO’s 2026 Draft Forecasting Assumption Update Inputs and Assumptions Workbook.
5.5.2 Interpretation of costs from SLCOE and LCOE
Neither SLCOE or LCOE estimates guarantee a future wholesale electricity generation price outcome. They indicate the breakeven price required for investors to make a return on their portfolio or individual investments respectively. SLCOE is the more accurate indicator of the two because it includes the integration costs and operational requirements of the entire system whereas LCOE does not with the much simpler calculation only including a limited set of standard cost and technical inputs
Neither of these cost estimates guarantee future electricity prices. Changes in electricity prices are also subject to17:
•
Supply-demand imbalance as a result of too much or too little deployment relative to demand growth and retirements.
•
Fuel price and weather volatility.
•
The level of competition amongst suppliers.
These additional drivers of price formation can lead to prices significantly lower or higher than the underlying cost of the system and can take many years to correct due to the long lead times for capacity deployment.
17 Retailers also have hedging costs, market fees and ancillary services costs associated with their purchase of wholesale electricity. However, we do not go into those topics here as we are focussed with the main underlying drivers of generation price changes.
56 | CSIRO Australia’s National Science Agency
Lead times are impacted by many factors such as the maturity of technology, approval and
development processes, general uncertainty, contract markets and confidence in government
policy directions. Construction times also vary significantly between technologies.
Given this background the SLCOE and LCOE data in GenCost are an indicator of the minimum price
needed for investors to enter the market of their own accord ignoring any other uncertainties or
external influences. However, should current or future electricity prices be much lower or higher
than the indicated breakeven SLCOE, this outcome is likely to have been caused by excess or
insufficient new capacity in the market whose root causes may span several years.
5.5.3 SLCOE results and the ISP
The SLCOE results will cover some similar ground to the Integrated System Plan (ISP) in that they
present a future generation mix for the NEM. Where results differ, GenCost advises that the ISP
should be given greater weight. The SLCOE results are based on a simplified electricity model. The
results are designed to be reasonably accurate but, by design, are substantially less sophisticated
than the multi-model state-of-the-art framework deployed in the ISP.
5.5.4 2030 LCOE results and future prices
Electricity futures prices for baseload and peak evening energy in New South Wales (NSW),
Victoria (VIC), Queensland (QLD) and South Australia (SA) are shown Figure 5-1.
Figure 5-1 Baseload and Peak evening electricity futures prices by state, nominal $/MWh
Baseload electricity futures prices are around $79/MWh on average for the next 3 years with a
rising trend, towards an average $85/MWh by 2029. For context, the AEMC (2025) projected
wholesale generation costs of just over $90/MWh in 2030 indicating reasonable alignment
between the futures market and available modelling results.
In Figure 5-2, technologies are grouped into peaking technologies and bulk energy technologies.
Peaking technologies have a capacity factor of 20% and this aligns well with the Peak evening
contract which applies 16:00 hours to 21:00 hours (AEST) Monday – Sunday18 representing a load
18 australian-electricity-fact-sheet.pdf
0
20
40
60
80
100
120
2027 2028 2029
$/MWh
NSW VIC QLD SA
Baseload
0
50
100
150
200
250
300
2027 2028 2029
$/MWh
NSW VIC QLD SA
Peak evening
GenCost 2025-26 | 57
factor of 21% of the year. Trades in Peak evening futures contracts are much lower and so there is a lower level of confidence in this futures price.
Bulk energy technologies might have an availability of up to just over 90% but often have much lower capacity factors in practice due to the lack of market loads requiring a 90% load factor. The Baseload electricity futures price is for a 100% or constant load factor throughout the year and so is not a perfect match but is the closest of available futures prices. The futures prices shown in Figure 5-2 are the lowest and highest of all states over the next three years. They have also been adjusted for inflation to make them consistent with the LCOE data which is in real terms.
LCOE excludes integration costs. See footnote previous page for futures price definitions
Figure 5-2 Calculated 2030 LCOE of peaking and bulk energy technologies compared to Baseload and Peak evening electricity futures price ranges
The comparison in Figure 5-2 indicates that peaking generation technologies are currently sitting at the high end of the Peak evening futures prices. This is consistent with observations that batteries have started to compete with traditional gas generation in this category, with the significant new capacity added resulting in reduced peak evening prices19. A four hour duration battery currently has a lower cost than a large open cycle gas turbine and this competitive advantage is expected to widen in the next two years (Figure 5-3).
Most bulk energy generation technologies are higher cost than the Baseload electricity futures price. This is an academic observation because most of these technologies could not be deployed within the next 4 years due to their required development and construction times.
19 https://www.aemo.com.au/energy-systems/major-publications/quarterly-energy-dynamics-qed
01002003004005006007002025 $/MWh2030 LCOE rangeFutures price lowFutures price highPeaking technologies Bulk energy technologies Variable technologies
58 | CSIRO Australia’s National Science Agency
Solar PV and onshore wind are almost exclusively the only generation technologies sufficiently advanced in the development pipeline to be deployed before 2030. Their deployment is being supported by existing flexible gas and coal together with new battery, pumped hydro and transmission capacity. Given the Baseload electricity futures price is an all-day, all-year price it represents the expected total annual generation cost of this combination of technologies. Transmission costs, which are recovered outside of the electricity system increase to 2030 but from a low base (AEMC 2025).
Figure 5-3 Historical and two year projection of capital cost of 4 hour duration battery and large open cycle gas turbine
5.5.5 2050 SLCOE results
Generation mix
By 2050, we assume all current generation is retired and only currently existing or committed hydro, pumped hydro and transmission remains. The existence of any rooftop solar PV, home batteries and vehicle-to-grid batteries are aligned to AEMO’s Step Change scenario assumptions.
The three technology groups considered in the analysis, in addition to mature renewables, are floating and fixed offshore wind, carbon capture and storage (CCS) applied to either coal or gas generation and large-scale and small modular reactor nuclear. However, to be eligible to deploy these technologies at the costs published in Appendix B of this report, each of these three technologies must first build a minimum of two projects at a cost premium consistent with the discussion of first-of-a-kind premiums in Section 2 of this report. The premiums applied are consistent with those presented in Table 2-1. In other words, there is an initial additional cost which must be paid to establish the required workforce, skills and supply chains when commencing a program of building technologies that Australia has not previously deployed.
0500100015002000250030002019202020212022202320242025202620272025 $/kW4 hour batteryLarge open cycle gas turbine
GenCost 2025-26 | 59
Mature renewables includes solar PV, onshore wind and hydro. FOAK – first of a kind. CCS – carbon capture and storage.
Figure 5-4 The projected generation mix in 2050 by emissions intensity target and allowed technology.
The least cost generation mix in 2050 by emission intensity target for either mature technologies only or mature technologies plus one of three technology groups not previously deployed in Australia are shown in Figure 5-4. In NoProgressToNetZero, with an emissions intensity 0.20tCO2e/MWh, the 2050 technology mix would include 70% renewables or slightly lower in those scenarios where the FOAK technologies are included. The remaining generation is from unabated fossil fuels which, in the past, GenCost has found are only slightly higher cost or similar cost to firmed renewables depending on prevailing fossil fuel prices. As such, it is not surprising that, where emission reduction targets are only mild, some fossil fuels will be deployed, even as a minority of generation.
There is an inconsistency in fossil assumptions for the NoProgressToNetZero emission intensity scenario. The fossil fuel prices are sourced from AEMO’s Input and Assumptions workbook where the scenarios assume that both Australia and the world are targeting substantial emissions reduction. In those scenarios, global and local fossil fuel demand is declining, keeping fossil fuel prices stable or in some cases falling. The NoProgressToNetZero emission intensity scenario is not consistent with those scenarios. 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.
0%10%20%30%40%50%60%70%80%90%100%Mature tech. only+ offshore wind+ CCS+ nuclearMature tech. only+ offshore wind+ CCS+ nuclearMature tech. only+ offshore wind+ CCS+ nuclearMature tech. only+ offshore wind+ CCS+ nuclearMature tech. only+ offshore wind+ CCS+ nuclearNoProgressToNetZeroWeakNetZeroModerateNetZeroStrongNetZeroVeryStrongNetZeroCoalNatural gasHydrogenMature renewablesNuclearCCSOffshore wind
60 | CSIRO Australia’s National Science Agency
While the model is free to do so, the least cost solution excludes any additional offshore wind, CCS or nuclear beyond the required initial FOAK projects in the NoProgressToNetZero emission intensity scenario.
In WeakNetZero (at 0.10tCO2e/MWh) the gas share of generation is decreased and there is no expansion of offshore wind, CCS or nuclear beyond the FOAK projects. Solar PV and onshore wind gain all of the market share lost by gas and coal relative to NoProgressToNetZero. In ModerateNetZero (at 0.05tCO2e/MWh), the same trend occurs with solar PV and onshore wind replacing natural gas with no competition from other generation sources.
In StrongNetZero (at 0.02tCO2e/MWh) the trend changes. Gas is too emission intensive to continue providing 7% of generation that it did under ModerateNetZero. However, gas cannot be cost effectively replaced with only solar PV and onshore wind because this would increase storage costs. Instead, the model deploys an alternative flexible generation technology in the form of hydrogen generation. The model does not deploy additional FOAK capacity beyond the required initial projects.
When the emission intensity target is set to zero in VeryStrongNetZero, the hydrogen share is increased from around 4% to 7% and natural gas decreases to less than 1% (this scenario allows for less than a million tonnes of residual emissions). Again, the model does not deploy additional FOAK capacity beyond the required initial projects.
Cost
The SLCOE results for all scenarios are shown in Figure 5-5. The SLCOE results for 2050 only allows for existing long-lived transmission and hydro technology, assuming all other large scale generation technology that exists today has retired20. For this reason, the 2050 estimates overstate the average cost of generation in 2050 by assuming there will be no significant existing generation capacity, but the estimates can be considered an upper bound on average costs21.
The SLCOE results show that deploying mature technology only (solar PV, wind, gas and storage) is the least cost generation mix in 2050 for all emission intensity levels modelled. For the FOAK technologies, the CCS and offshore wind scenarios have very similar costs with CCS being only slightly higher cost than offshore wind. Nuclear is consistently the highest cost. These outcomes are based on average costs. Offshore wind has a much wider cost uncertainty range and so could perform better under alternative cost scenarios not explored.
20 This is not strictly true because most solar PV built now or in the last few years will remain operating in 2050 due to their 30 year life.
21 In practice, the partial existing capacity that would normally be available to meet 2050 demand will be developed and paid for through generation in the decades leading up to 2050 and building out this investment pathway is how a more commercial grade electricity system model would estimate electricity costs over time. However, in this simplified modelling approach we only include selected long-lived existing resources and other resources needed to meet demand in 2050 are built in 2050 and paid for in the decades that follow (through amortisation).
GenCost 2025-26 | 61
Figure 5-5 The projected SLCOE in 2050 in the NEM by emission intensity target and technology allowed
There is a clear trend that decreasing emissions increases 2050 electricity costs but lower cost emission intensity scenarios may increase costs elsewhere in the economy and so should be interpreted with caution. We identify the efficient level of emissions to reach net zero further below. The rising trend reflects that as the emissions intensity declines, more zero emissions technology must be deployed, supported by more storage and more transmission as well as more expensive fuels (such as hydrogen) in some cases (StrongNetZero and VeryStrongNetZero).
The breakdown of costs for each mature technology emissions intensity scenario is shown in Figure 5-6. The fuel cost falls as the gas share of generation declines but increases for the last two scenarios as hydrogen enters the generation mix with hydrogen fuel being higher cost than gas on an energy unit basis. Baseload fossil fuel costs also fall as they are removed from the generation mix as the emission intensity declines. Almost all other cost categories increase as the emissions intensity falls. However, the increase in connection costs, inverter-based resource costs and operating and maintenance costs are relatively minor. The biggest increases are in storage and transmission costs.
0204060801001201401601802002025 $/MWhMaturetechnology onlyMature +offshore windMature + CCSMature + nuclear
62 | CSIRO Australia’s National Science Agency
VRE-– variable renewable generation; IBR - Inverter-based resources cost such as deployment of synchronous condensers and grid-forming batteries; REZ – renewable energy zone. O&M – operating and maintenance costs
Figure 5-6 The breakdown of SLCOE in 2050 in the NEM by cost component for mature technology only scenario
As discussed in the scenario descriptions, in theory the highest of the projected costs in 2050 need only be experienced if lower cost abatement is not available elsewhere in the economy outside of the electricity sector since there is no specific requirement for the electricity sector to eliminate all emissions. To find the efficient level of electricity sector abatement, the marginal (incremental) cost of abatement of each emissions intensity level was calculated relative to the next highest emission intensity scenario (Figure 5-7). The cost of abatement in the electricity sector can be compared to the expected cost of abatement across the whole economy to reach net zero which is published by Infrastructure Australia22. They publish a cost of abatement range in 2050 of $304 to $497/tCO2e (adjusted to 2025 dollars).
The analysis shows that it is not efficient for the electricity sector to implement the VeryStrongNetZero emission intensity scenario which almost eliminates emissions from the sector (with less than a million tonnes remaining) because it is expected that there will be lower cost abatement in other sectors of the economy. If the whole of economy cost of abatement is in the high end of the range, then it would be efficient for the electricity sector to implement StrongNetZero. On the other hand, if the whole of economy cost of abatement is at the lower end of the expected range then the StrongNetZero emission intensity scenario is not efficient as it would incur costs that are higher than abatement options in other sectors of the economy.
At the lowest end of the whole of economy cost range, it would be efficient for the electricity sector to target an emissions intensity of 0.05tCO2e/MWh in the ModerateNetZero scenario or slightly lower that given its marginal cost of abatement is below the whole of economy range.
22 https://www.infrastructureaustralia.gov.au/publications/valuing-emissions-economic-analysis
0204060801001201401601802025 $/MWhOther transmissionREZ transmissionFuelConnectionIBR costO&MStorageBaseload fossil capitalPeaking capitalVRE capital
GenCost 2025-26 | 63
Figure 5-7 Average and marginal cost of abatement to achieve lower emissions intensity targets in 2050 compared to whole of economy abatement costs
In summary, examining the cost of abatement indicates that:
•
In a whole of economy effort to reach net zero by 2050, efficient level of emission abatement in the electricity sector is uncertain given the cost of abatement in the rest of the economy has a wide uncertainty range.
•
Based on the range of whole of economy abatement costs, the efficient range of emissions intensity of the electricity sector is 0.05tCO2e/MWh to 0.02tCO2e/MWh.
•
Achieving the electricity sector’s efficient role in whole of economy net zero abatement is projected to result in electricity costs $141/MWh to $152/MWh or an average of $146/MWh in the NEM inclusive of new transmission costs or $120/MWh to $130/MWh or an average of $125/MWh measured as wholesale generation costs only.
•
The average cost of implementing the efficient level of abatement in the electricity sector of 0.02tCO2e/MWh is $175/tCO2e. 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.
5.5.6 Alternative results for cost of high renewable systems in the literature
There is a surprising range of estimates of the cost of electricity from systems with a high share of weather dependent renewable generation. These studies have been separately reviewed in Graham (2025) where costs have been estimated in the range of below $70/MWh to over $1000/MWh. The review finds that, in each case where system average costs were reported to be high, the published modelling had excluded key resources required to keep costs low. Typical exclusions included: only one type of storage technology allowed, only one duration of storage
0100200300400500600700WeakNetZeroModerateNetZeroStrongNetZeroVeryStrongNetZero2025 $/tCO2eAverage costMarginal costWhole of economy cost highWhole of economy cost low
64 | CSIRO Australia’s National Science Agency
technology allowed, peaking technology not allowed and a narrow set or single type of renewable generation allowed. When resources are made available, renewable generation is lowest cost when sourced from both solar PV and wind in different locations and combined with multiple storage technologies and gas or hydrogen peaking plant. Graham (2025) concludes that the observed exclusions in other studies do not appear to be valid for the NEM which has all of these resources available.
5.5.7 2050 LCOE estimates
In addition to the SLCOE estimates provided in the previous section, the LCOE for individual technologies is also calculated and represents the breakeven price needed for each technology to achieve a reasonable return on investment. Appendix B includes the LCOE for additional years and technologies, however in this section we focus on a selected set of technologies for consistency with the SLCOE modelling.
Figure 5-8 shows the LCOE results for 2050. Here they have been compared to the electricity system cost range identified in the SLCOE analysis as being the potential efficient range for electricity sector abatement from ModerateNetZero to VeryStrongNetZero. By 2050, the costs of most technologies other than coal and gas have fallen compared to 2025. Solar PV and onshore wind (without integration costs) are lowest cost relative to the SLCOE estimates. The SLCOE analysis, which does include the renewable integration costs, indicates that solar PV and wind will deliver the majority of electricity supply (93%) supported by hydro, storage, transmission and either gas or hydrogen or a combination of both.
New black coal, while competitive at this SLCOE range is not relevant for deployment if the goal is to efficiently achieve net zero (that is, it would increase the cost of achieving net zero across the economy). New gas is also in the competitive range and unlike coal, the SLCOE analysis indicates it will play a role in achieving net zero emissions contributing a 3% to 7% share of generation.
Solar thermal is competitive relative to other technologies and inside the SLCOE range. However, given the need to access better solar resources which are further from load centres, solar thermal will be subject to additional transmission costs compared to coal, gas and nuclear which have not been directly accounted for. Based on Figure 5-6, additional transmission costs could add around $20/MWh. Some offshore wind is also in the competitive range however the modelling used the average cost of offshore wind. Lower range offshore wind costs appear to be competitive but were not explored in the modelling and so need more analysis.
Gas with CCS is the next most competitive after solar thermal and offshore wind. Large-scale nuclear is slightly higher in cost than gas with CCS. Black coal with CCS occupies a similar cost range to nuclear. Nuclear small modular reactors (SMRs) are the highest cost. Achieving the lower end of the nuclear SMR range requires that SMR is deployed globally in large enough capacity to bring down costs available to Australia. Lowest cost gas with CCS is subject to accessing gas supply at the lower end of the range assumed (see Appendix B for fuel cost assumptions). Coal, gas and nuclear technologies would all have to be successful in operating at 89% capacity factor to achieve the lower end of the cost range when historically coal, which has been the main baseload energy source in Australia’s largest states, has only achieved an average of around 60%.
GenCost 2025-26 | 65
LCOE excludes integration costs. SLCOE includes integration costs of the least cost technology mix which includes a combined solar PV and onshore wind share of 87% to 89%.
Figure 5-8 Calculated LCOE range by technology and SLCOE range for 2050
5.6 Overall implications for long term electricity generation cost trends
In 2025 calendar year the average NEM volume weighted generation price was estimated to be $104/MWh, down from the recent peak in 2022 of $189/MWh caused predominantly by high gas prices. Based on NEM electricity futures prices in the next three years (and supported by AEMC (2025) modelling), generation costs to 2030 are expected to fall further compared to 2025 to around $80-$90/MWh.
In the post-2030 period, net zero by 2050 is the most common type of climate change policy adopted globally and at the national, state and corporate level in Australia. By 2050, most generation capacity that exists today will be retired. The LCOE and SLCOE analysis in this report indicates that there are no new build technology options, fossil or non-fossil including integration costs, that cost less than $100/MWh (Figure 5-8). By 2050, new build coal generation costs are in the range of $107/MWh to $182/MWh. As such, even without the net zero by 2050 policies, generation prices are expected to increase above $100/MWh in the post 2030 period.
The cost of generation if the electricity sector makes an efficient contribution to achieving net zero by 2050 is $141/MWh to $152/MWh inclusive of all firming costs (an average of $146/MWh). This is based on a 2050 generation mix of large-scale solar PV and onshore wind of around 93% in total. Excluding transmission costs which are not recovered through the generation market, the net zero consistent generation cost is $120/MWh to $130/MWh or an average of $125/MWh.
01002003004002025 $/MWhLCOE rangeSLCOE lowSLCOE high
66 | CSIRO Australia’s National Science Agency
Global and local learning model
A.1 GALLM
The Global and Local Learning Models (GALLMs) for electricity (GALLME) and transport (GALLMT)
are described briefly here. More detail can be found in several publications (Hayward and Graham,
2017; Hayward and Graham, 2013; Hayward, Foster, Graham and Reedman, 2017).
A.1.1 Endogenous technology learning
Technology cost reductions due to ‘learning-by-doing’ were first observed in the 1930s for
aeroplane construction (Wright, 1936) and have since been observed and measured for a wide
range of technologies and processes (McDonald and Schrattenholzer, 2001). Cost reductions due
to this phenomenon are normally shown via the equation:
where IC is the unit investment cost at CC cumulative capacity and IC0 is the cost of the first unit at
CC0 cumulative capacity. The learning index b satisfies 0 < b < 1 and it determines the learning rate
which is calculated as:
(typically quoted as a percentage ranging from 0 to 50%) and the progress ratio is given by
PR=100-LR. All three quantities express a measure of the decline in unit cost with learning or
experience. This relationship states that for each doubling in cumulative capacity of a technology,
its investment cost will fall by the learning rate (Hayward & Graham, 2013). Learning rates can be
measured by examining the change in unit cost with cumulative capacity of a technology over
time.
Typically, emerging technologies have a higher learning rate (15–20%), which reduces once the
technology has at least a 5% market share and is considered to be at the intermediate stage (to
approximately 10%). Once a technology is considered mature, the learning rate tends to be 0–5%
(McDonald and Schrattenholzer, 2001). The transition between learning rates based on technology
uptake is illustrated in Apx Figure A.1.
𝐼𝐶 = 𝐼𝐶0 ×
𝐶𝐶
𝐶𝐶0
−𝑏
,
or equivalently log 𝐼𝐶 = log 𝐼𝐶0 − 𝑏(log 𝐶𝐶 − log 𝐶𝐶0 )
𝐿𝑅 = 100 × 1 − 2−𝑏
GenCost 2025-26 | 67
Apx Figure A.1 Schematic of changes in the learning rate as a technology progresses through its development stages after commercialisation
However, technologies that are modular and as a result can be used in a variety of applications tend to have a higher learning rate for longer (Wilson, 2012). This is the case for solar photovoltaics, batteries and historically for gas turbines.
Technologies are made up of components and different components can be at different levels of maturity and thus have different learning rates. Different parts of a technology can be developed and sold in different markets (global vs. regional/local) which can impact the relative cost reductions given each region will have a different level of demand for a technology.
A.1.2 The modelling framework
To project the future cost of a technology using experience curves, the future level of cumulative capacity/uptake needs to be known. However, this is dependent on the costs. The GALLM models solve this problem by simultaneously projecting both the cost and uptake of the technologies. The optimisation problem includes constraints such as government policies, demand for electricity or transport, capacity of existing technologies, exogenous costs such as for fossil fuels and limits on resources (e.g., rooftops for solar photovoltaics). The models have been divided into 13 regions and each region has unique assumptions and data for the above listed constraints. The regions have been based on Organisation for Economic Co-operation Development (OECD) regions (with some variation to look more closely at some countries of interest) and are Africa, Australia, China, Eastern Europe, Western Europe, Former Soviet Union, India, Japan, Latin America, Middle East, North America, OECD Pacific, Rest of Asia and Pacific.
The objective of the model is to minimise the total system costs while meeting demand and all constraints. The model is solved as a mixed integer linear program. The experience curves are segmented into step functions and the location on the experience curves (i.e., cost vs. cumulative
68 | CSIRO Australia’s National Science Agency
capacity) is determined at each time step. See Hayward and Graham (2013) and Hayward et al. (2017) for more information. Both models run from the year 2006 to 2100. However, results are only reported from the present year to 2055.
A.1.3 Offshore wind
Offshore wind has been divided into fixed and floating foundation technologies. IRENA (2024) and Stehly and Duffy (2021) provided a breakdown of the cost of all components of both fixed and floating offshore wind, which allowed us to separate out the cost of the foundations from the remainder of the cost components. This division in costs was then applied to the current Australian costs from GHD (2026) resulting in the values as shown in Apx Table A.1.
Apx Table A.1 Cost breakdown of offshore wind Cost component Fixed offshore wind ($/kW) Floating offshore wind ($/kW) Foundation 597 2393
Remainder of cost
4065
4065 Total cost 4662 6459
The learning of all offshore wind components (i.e., “Remainder of cost” components) except for the foundations are shared among both offshore wind technologies. The floating foundations used in floating offshore wind have a learning rate, but the fixed foundations used in fixed offshore wind have no learning rate.
GenCost 2025-26 | 69
Data tables
The following tables provide data behind the figures presented in this document.
The year 2025 is mostly sourced from GHD (2026) and is aligned to July which represents either the middle of that calendar year or the beginning of the 2025-26 financial year. All projections are in real 2025 dollars.
As discussed in Section 2, the data is not intended to include FOAK costs. Therefore, for technologies not recently constructed in Australia, the cost of the first and some subsequent deployments may be higher than shown in this appendix. Section 2 includes suggested FOAK premiums.
GHD provide data for adjusting costs for different locations in the NEM and this data is also published by AEMO. Site conditions will also impact costs to varying degrees, depending on the technology.
All capital costs are for the alternating current power rating of the equipment with the exception of rooftop solar which is on a direct current basis. Power is also on a net basis after auxiliary loads. Capital costs are before any subsidies that may be available.
70 | CSIRO Australia’s National Science Agency
Apx Table B.1 Current and projected generation technology capital costs under the Current policies scenario
Black coal Black coal with CCS Brown coal Gas combined cycle Gas open cycle (small) Gas open cycle (large) Gas with CCS Gas reciprocating Hydrogen reciprocating Biomass (small scale) Biomass with CCS (large scale) Large scale solar PV Rooftop solar panels Solar thermal (14hrs) Wind Offshore wind fixed Offshore wind floating Wave Nuclear SMR Tidal /ocean current Fuel cell Nuclear large-scale
$/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW
2025
6946
12941
10725
2497
2914
1764
6962
2022
2648
9016
25712
1621
1216
6493
3248
5433
8325
15842
28009
12420
7000
10332 2026 6946 12941 10725 2640 3083 1867 7105 2022 2648 9016 25712 1516 1191 6453 3248 5380 7261 15842 26351 12420 7000 10332
2027
6584
12465
10108
2783
3253
1970
7248
2010
2594
8387
24850
1443
1171
6435
2984
5342
6559
14835
24692
11500
6613
10000 2028 6435 12284 9852 2658 3023 1831 6906 2009 2573 8287 24430 1401 1156 6424 2866 5320 6224 14356 23034 11066 6428 9877
2029
6296
12118
9613
2539
2814
1704
6580
2009
2554
8316
24017
1388
1145
6395
2755
5306
6202
13893
21452
10649
6248
9764 2030 6162 11961 9385 2426 2624 1589 6269 2010 2537 8349 23611 1378 1135 6358 2650 5300 6194 13444 20806 10247 6073 9658
2031
6031
11805
9161
2317
2425
1468
5973
2011
2519
8381
23211
1358
1127
6327
2553
5300
6195
13010
20325
9860
5903
9552 2032 5899 11644 8938 2213 2242 1357 5691 2011 2501 8411 22819 1338 1121 6308 2464 5307 6201 12590 19830 9488 5738 9442
2033
5769
11481
8717
2114
2073
1255
5422
2010
2482
8439
22433
1319
1117
6295
2386
5311
6206
12183
19915
9130
5577
9330 2034 5641 11320 8501 2020 1919 1162 5166 2010 2463 8466 22054 1300 1112 6287 2323 5314 6209 11790 20000 8786 5421 9220
2035
5564
11230
8371
1929
1776
1076
4922
2012
2454
8494
21681
1281
1107
6284
2309
5316
6211
11409
20085
8454
5269
9159 2036 5536 11210 8325 1916 1747 1058 4873 2017 2453 8522 21771 1262 1101 6284 2275 5317 6207 11489 20171 8493 5283 9149
2037
5558
11260
8362
1922
1752
1061
4888
2024
2462
8551
21865
1244
1098
6220
2268
5316
6189
11502
20258
8499
5291
9189 2038 5581 11310 8400 1928 1757 1064 4904 2031 2471 8581 21960 1226 1094 6115 2262 5316 6157 11515 19900 8504 5294 9229
2039
5603
11361
8437
1934
1762
1067
4919
2038
2480
8610
22056
1208
1092
5979
2257
5315
6118
11528
18488
8510
5295
9269 2040 5624 11408 8472 1939 1767 1070 4933 2045 2488 8637 22144 1191 1090 5873 2253 5315 6093 11539 16682 8516 5284 9306
2041
5642
11450
8503
1944
1770
1072
4944
2050
2494
8659
22222
1174
1088
5791
2249
5316
6082
11550
15308
8522
5276
9339 2042 5659 11487 8531 1948 1773 1074 4954 2055 2500 8679 22292 1157 1086 5726 2246 5317 6084 11560 14974 8528 5274 9368
2043
5677
11524
8558
1952
1776
1076
4963
2060
2506
8698
22362
1140
1084
5684
2242
5320
6088
11569
15021
8534
5280
9398 2044 5695 11562 8586 1955 1780 1078 4973 2065 2512 8717 22433 1123 1083 5622 2236 5322 6091 11579 15068 8540 5280 9428
2045
5713
11600
8615
1959
1783
1079
4982
2070
2518
8737
22504
1107
1082
5546
2232
5325
6095
11589
15116
8545
5265
9457 2046 5730 11638 8643 1963 1786 1081 4992 2075 2524 8757 22575 1091 1081 5465 2227 5328 6098 11598 15164 8551 5242 9487
2047
5747
11676
8671
1967
1789
1083
5002
2080
2530
8776
22647
1075
1080
5414
2222
5332
6102
11608
15212
8557
5207
9517 2048 5762 11715 8700 1971 1792 1085 5012 2085 2537 8796 22719 1060 1079 5390 2216 5335 6106 11618 15260 8563 5151 9548
2049
5777
11754
8729
1974
1795
1087
5022
2090
2543
8816
22792
1044
1077
5380
2210
5338
6110
11628
15309
8569
5082
9578 2050 5791 11790 8756 1978 1798 1089 5031 2095 2548 8835 22860 1029 1077 5376 2207 5342 6114 11637 15355 8576 5026 9607
2051
5804
11825
8782
1981
1801
1090
5039
2099
2554
8852
22925
1014
1077
5373
2204
5345
6117
11646
15399
8582
4992
9634 2052 5815 11858 8806 1984 1803 1092 5046 2103 2558 8867 22986 999 1077 5369 2201 5348 6121 11654 15439 8588 4975 9660
2053
5827
11890
8830
1987
1806
1093
5053
2107
2563
8882
23047
985
1077
5358
2198
5351
6124
11662
15480
8594
4953
9685 2054 5838 11923 8855 1990 1808 1095 5060 2111 2568 8898 23108 971 1077 5347 2194 5355 6129 11671 15522 8600 4935 9711
2055
5843
11940
8867
1991
1809
1095
5064
2113
2571
8906
23138
957
1077
5341
2192
5357
6131
11675
15542
8603
4928
9724
GenCost 2025-26 | 71
Apx Table B.2 Current and projected generation technology capital costs under the Global NZE by 2050 scenario
Black coal Black coal with CCS Brown coal Gas combined cycle Gas open cycle (small) Gas open cycle (large) Gas with CCS Gas reciprocating Hydrogen reciprocating Biomass (small scale) Biomass with CCS (large scale) Large scale solar PV Rooftop solar panels Solar thermal (14hrs) Wind Offshore wind fixed Offshore wind floating Wave Nuclear SMR Tidal /ocean current Fuel cell Nuclear large-scale
$/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW
2025
6946
12941
10725
2497
2914
1764
6962
2022
2648
9016
25712
1621
1216
6493
3248
5433
8325
15842
28009
12420
7000
10332 2026 6946 12941 10725 2640 3083 1867 7105 2022 2648 9016 25712 1419 1174 6345 3248 5115 7124 15842 26351 12420 7000 10332
2027
6617
12514
10156
2783
3253
1970
7248
2016
2602
8702
24937
1237
1137
6228
2971
4816
6049
14923
24692
11577
6616
10046 2028 6498 12387 9946 2665 3030 1835 6900 2022 2590 8578 24558 1083 1107 6148 2849 4544 5145 14484 23034 11177 6433 9967
2029
6396
12291
9765
2552
2827
1712
6568
2030
2581
8469
24184
950
1080
6084
2735
4293
4382
14057
21716
10791
6254
9912 2030 6296 12196 9588 2443 2642 1600 6252 2039 2574 8408 23817 847 1055 6037 2629 4133 3917 13643 21273 10418 6080 9858
2031
6188
12083
9399
2340
2447
1481
5951
2046
2563
8364
23455
783
1025
6002
2529
4059
3705
13242
18574
10058
5911
9790 2032 6074 11952 9200 2240 2267 1373 5665 2050 2550 8368 23098 757 1004 5981 2436 4064 3707 12852 15440 9711 5747 9708
2033
5957
11815
8999
2145
2101
1272
5393
2054
2536
8367
22747
751
977
5943
2355
4070
3709
12474
12661
9375
5587
9620 2034 5843 11680 8803 2054 1949 1180 5133 2058 2522 8365 22401 750 957 5892 2289 4076 3713 12106 12099 9052 5432 9533
2035
5782
11517
8695
1967
1808
1095
4886
2064
2517
8341
22061
747
926
5824
2273
4083
3717
11750
11903
8739
5281
9499 2036 5772 11416 8674 1957 1781 1078 4741 2073 2522 8327 22115 725 901 5770 2234 4090 3722 11643 11919 8583 5304 9516
2037
5814
11383
8739
1966
1788
1083
4658
2085
2536
8330
22168
699
881
5728
2225
4098
3727
11669
11955
8572
5323
9585 2038 5857 11446 8806 1975 1796 1088 4660 2097 2551 8348 22310 676 871 5700 2216 4107 3733 11695 12043 8561 5348 9656
2039
5900
11504
8873
1985
1804
1092
4657
2109
2566
8359
22448
673
862
5678
2214
4117
3740
11722
12131
8572
5360
9727 2040 5941 11561 8937 1994 1811 1097 4654 2120 2579 8385 22582 672 855 5662 2214 4126 3748 11747 12216 8588 5366 9795
2041
5981
11621
8998
2002
1818
1101
4657
2131
2592
8416
22714
669
836
5650
2213
4137
3757
11772
12297
8214
5376
9859 2042 6018 11694 9055 2009 1824 1104 4672 2141 2604 8451 22852 666 817 5642 2209 4146 3765 11795 12373 7819 5379 9920
2043
6055
11770
9114
2016
1830
1108
4689
2151
2616
8487
22994
663
795
5637
2205
4157
3773
11819
12450
7423
5386
9982 2044 6093 11843 9173 2023 1836 1112 4704 2160 2628 8524 23135 662 784 5635 2202 4167 3782 11843 12527 7410 5336 10044
2045
6131
11915
9232
2031
1842
1115
4717
2171
2640
8561
23276
661
778
5637
2199
4177
3791
11867
12606
7409
5290
10107 2046 6169 11987 9292 2038 1848 1119 4729 2181 2653 8598 23417 660 773 5643 2197 4188 3800 11891 12685 7408 5198 10171
2047
6208
12061
9353
2046
1855
1123
4743
2191
2665
8636
23560
656
767
5658
2194
4199
3809
11916
12765
7413
5133
10235 2048 6247 12136 9414 2054 1861 1127 4757 2201 2678 8674 23705 647 756 5676 2190 4210 3818 11940 12845 7428 5053 10299
2049
6287
12210
9476
2061
1867
1131
4771
2212
2691
8712
23851
631
742
5700
2185
4221
3826
11965
12927
7442
5007
10365 2050 6326 12282 9537 2069 1874 1134 4782 2222 2703 8749 23992 616 731 5723 2180 4231 3834 11990 13007 7457 4970 10429
2051
6364
12349
9597
2076
1879
1138
4791
2232
2715
8785
24128
606
723
5742
2177
4241
3843
12014
13085
7471
4956
10491 2052 6401 12416 9655 2083 1885 1141 4799 2241 2726 8819 24262 603 723 5755 2176 4252 3852 12038 13161 7486 4943 10553
2053
6439
12483
9714
2090
1891
1145
4808
2251
2738
8854
24396
602
723
5746
2176
4262
3860
12062
13239
7501
4934
10615 2054 6476 12553 9773 2096 1896 1148 4819 2260 2750 8889 24534 602 725 5724 2174 4273 3869 12087 13316 7516 4908 10677
2055
6495
12588
9803
2100
1899
1150
4824
2265
2756
8907
24603
602
726
5705
2173
4278
3873
12099
13355
7524
4894
10708
72 | CSIRO Australia’s National Science Agency
Apx Table B.3 Current and projected generation technology capital costs under the Global NZE post 2050 scenario
Black coal Black coal with CCS Brown coal Gas combined cycle Gas open cycle (small) Gas open cycle (large) Gas with CCS Gas reciprocating Hydrogen reciprocating Biomass (small scale) Biomass with CCS (large scale) Large scale solar PV Rooftop solar panels Solar thermal (14hrs) Wind Offshore wind fixed Offshore wind floating Wave Nuclear (SMR) Tidal /ocean current Fuel cell Nuclear large-scale
$/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW $/kW
2025
6946
12941
10725
2497
2914
1764
6962
2022
2648
9016
25712
1621
1216
6493
3248
5433
8325
15842
28009
12420
7000
10332 2026 6946 12941 10725 2640 3083 1867 7105 2022 2648 9016 25712 1566 1194 6454 3248 5396 7787 15842 26351 12420 7000 10332
2027
6596
12472
10121
2783
3253
1970
7248
2012
2596
8704
24937
1515
1176
6441
2981
5366
7281
14906
24692
11561
6613
10013 2028 6456 12303 9879 2660 3026 1832 6904 2012 2578 8572 24558 1471 1162 6458 2863 5344 6817 14460 23034 11155 6428 9902
2029
6329
12156
9658
2543
2818
1706
6576
2015
2562
8451
24184
1430
1151
6485
2751
5326
6386
14026
21487
10762
6248
9808 2030 6207 12015 9446 2431 2630 1592 6264 2019 2548 8379 23817 1392 1137 6511 2647 5292 6092 13606 20972 10384 6073 9718
2031
6084
11869
9233
2324
2432
1472
5967
2021
2532
8332
23455
1358
1130
6524
2549
5246
5927
13198
20477
10018
5903
9624 2032 5957 11713 9016 2222 2250 1362 5684 2022 2515 8327 23098 1325 1123 6530 2460 5202 5894 12802 19997 9666 5738 9522
2033
5832
11556
8802
2124
2082
1261
5414
2023
2498
8323
22747
1293
1116
6530
2381
5164
5865
12418
20101
9326
5577
9418 2034 5710 11402 8593 2030 1928 1167 5157 2024 2481 8342 22401 1261 1109 6538 2318 5116 5788 12046 18260 8998 5421 9315
2035
5639
11321
8471
1941
1786
1081
4912
2028
2473
8352
22061
1230
1103
6541
2303
5053
5698
11685
16275
8681
5269
9264 2036 5619 11312 8434 1929 1758 1064 4865 2034 2475 8362 22177 1200 1096 6447 2270 4996 5614 11532 14280 8524 5284 9264
2037
5650
11375
8481
1936
1764
1068
4882
2043
2485
8373
22297
1171
1089
6303
2266
4957
5584
11548
14222
8533
5300
9314 2038 5680 11439 8529 1943 1770 1071 4901 2052 2496 8392 22418 1143 1082 6134 2261 4937 5570 11564 14294 8541 5292 9365
2039
5711
11497
8577
1950
1776
1075
4914
2061
2507
8409
22535
1115
1075
6059
2258
4927
5487
11581
14366
8549
5061
9416 2040 5740 11552 8622 1957 1781 1078 4925 2069 2517 8428 22644 1088 1069 6007 2251 4929 5414 11596 13994 8558 4817 9464
2041
5767
11602
8663
1963
1786
1081
4934
2077
2526
8442
22744
1061
1062
5963
2246
4933
5341
11610
13553
8566
4587
9508 2042 5792 11653 8702 1968 1790 1084 4947 2083 2534 8458 22841 1036 1056 5905 2240 4941 5350 11623 13103 8574 4518 9548
2043
5816
11704
8740
1973
1794
1086
4959
2090
2542
8472
22938
1010
1049
5834
2241
4919
5333
11636
13047
8583
4412
9589 2044 5841 11756 8779 1978 1798 1089 4972 2096 2550 8455 23036 986 1043 5748 2242 4872 5297 11650 12741 8591 4229 9630
2045
5866
11808
8818
1983
1802
1091
4985
2103
2558
8422
23135
962
1036
5657
2241
4812
5249
11663
12375
8600
4096
9671 2046 5892 11861 8857 1988 1807 1094 4998 2110 2567 8376 23235 938 1030 5604 2238 4775 5221 11676 12052 8609 4005 9713
2047
5917
11914
8897
1993
1811
1096
5010
2117
2575
8341
23335
915
1023
5574
2232
4752
5204
11690
12045
8617
3985
9754 2048 5942 11967 8937 1998 1815 1099 5023 2124 2583 8299 23436 903 1017 5563 2227 4734 5191 11703 12097 8626 3981 9797
2049
5968
12021
8977
2003
1819
1102
5036
2130
2592
8274
23537
888
1010
5554
2219
4721
5182
11717
12149
8635
3978
9839 2050 5993 12073 9015 2008 1823 1104 5049 2137 2600 8272 23635 873 1004 5548 2213 4721 5183 11730 12200 8644 3982 9880
2051
6017
12123
9053
2013
1827
1106
5060
2143
2607
8294
23729
860
1000
5536
2207
4727
5189
11742
12248
8652
3988
9919 2052 6040 12171 9088 2017 1831 1108 5071 2149 2614 8315 23819 851 1000 5523 2203 4730 5194 11754 12295 8661 3992 9957
2053
6063
12219
9125
2021
1834
1110
5081
2155
2621
8336
23910
844
1000
5520
2198
4718
5186
11767
12326
8670
3993
9995 2054 6086 12257 9161 2025 1838 1113 5082 2161 2629 8357 23991 836 1001 5530 2191 4697 5170 11779 12345 8679 3988 10033
2055
6097
12276
9179
2027
1839
1114
5083
2164
2632
8367
24032
829
1001
5537
2188
4681
5158
11785
12347
8684
3984
10052
GenCost 2025-26 | 73
Apx Table B.4 One- and two-hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW)
Battery storage (1 hr)
Battery storage (2 hrs)
Total
Battery
BOP
Total
Battery
BOP
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
2025
778
778
778
310
310
311
467
467
467
525
525
525
290
290
290
235
235
235 2026 743 733 723 301 298 295 443 436 428
503 497 491 281 278 276 223 219 215
2027
707
692
678
285
284
283
422
408
395
478
470
463
266
265
265
212
205
198 2028 678 657 637 272 272 273 406 385 364
458 447 438 254 254 255 204 193 183
2029
661
631
601
264
264
264
397
367
337
446
430
416
247
246
246
199
184
169 2030 653 620 588 256 255 256 397 365 333
438 421 406 239 238 238 199 183 167
2031
645
608
575
248
246
248
396
362
327
430
411
395
232
230
231
199
182
164 2032 636 596 561 241 238 240 396 359 322
423 402 385 224 222 223 199 180 162
2033
628
585
548
233
229
232
395
356
316
415
392
375
217
214
216
198
179
159 2034 620 575 536 226 222 224 394 353 311
408 383 365 210 206 209 198 177 156
2035
613
564
523
219
214
217
394
350
306
401
375
356
204
199
202
197
176
153 2036 605 554 511 212 207 210 393 347 301
394 366 347 197 192 196 197 174 151
2037
598
544
499
205
199
204
393
345
296
388
358
338
191
186
190
197
173
148 2038 591 535 488 199 193 197 392 342 291
381 350 329 185 179 183 196 171 146
2039
584
525
477
193
186
191
392
339
286
375
343
321
179
173
178
196
170
143 2040 582 521 472 191 185 190 391 337 281
373 340 318 178 172 177 196 168 141
2041
580
517
467
190
183
190
390
334
277
372
337
315
176
170
177
195
167
138 2042 578 513 462 188 182 190 389 331 272
370 335 312 175 169 176 195 166 136
2043
576
509
457
187
181
190
389
328
267
369
333
310
174
169
176
194
164
134 2044 574 506 452 187 181 190 388 325 262
367 331 308 173 168 176 194 163 131
2045
573
502
447
186
180
190
387
322
258
366
329
305
173
167
176
193
161
129 2046 571 499 443 185 180 190 386 319 253
365 327 303 172 167 176 193 160 127
2047
570
496
439
185
179
190
385
317
249
364
325
301
172
167
176
193
158
125 2048 569 493 435 185 179 190 384 314 245
364 323 299 171 166 176 192 157 122
2049
568
490
431
184
179
190
383
311
240
363
322
297
171
166
177
192
156
120 2050 567 487 427 184 179 190 383 308 236
362 320 295 171 166 177 191 154 118
2051
565
484
423
184
179
190
382
306
232
361
319
293
171
166
177
191
153
116 2052 564 481 419 184 178 191 381 303 228 361 317 291 170 166 177 190 151 114
2053
563
478
415
183
178
191
380
300
224
360
316
289
170
166
177
190
150
112 2054 562 476 411 183 178 191 379 297 220 359 314 288 170 166 178 189 149 110
2055
569
485
421
183
178
191
386
306
230
363
319
293
170
166
178
193
153
115
74 | CSIRO Australia’s National Science Agency
Apx Table B.5 Four- and eight-hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW)
Battery storage (4 hrs)
Battery storage (8 hrs)
Total
Battery
BOP
Total
Battery
BOP
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
2025
385
385
385
265
265
265
120
120
120
308
308
308
245
245
245
63
63
63 2026 370 365 361 256 254 251 114 112 110
296 293 290 236 234 232 60 59 58
2027
350
346
342
242
242
241
108
105
101
280
278
276
224
223
223
57
55
53 2028 335 330 326 231 232 232 104 99 93
268 265 263 213 214 214 55 52 49
2029
326
318
311
225
224
224
102
94
86
261
256
252
208
207
207
53
49
45 2030 319 310 302 218 217 217 101 93 85
254 249 245 201 200 201 53 49 45
2031
312
302
294
211
209
210
101
92
84
248
241
238
195
193
194
53
48
44 2032 305 293 286 204 202 203 101 92 82
241 234 231 188 186 188 53 48 43
2033
299
286
278
198
195
197
101
91
81
235
227
224
182
180
182
53
48
42 2034 292 278 270 191 188 190 101 90 79
229 220 217 177 173 176 53 47 42
2035
286
271
262
185
181
184
100
89
78
223
214
211
171
167
170
53
47
41 2036 280 263 255 179 175 178 100 89 77
218 208 204 165 161 164 52 46 40
2037
274
257
248
174
169
172
100
88
75
212
202
198
160
156
159
52
46
39 2038 268 250 241 168 163 167 100 87 74
207 196 192 155 150 154 52 46 39
2039
262
243
234
163
157
161
100
86
73
202
190
187
150
145
149
52
45
38 2040 261 241 232 161 156 161 99 86 72
201 188 186 149 144 148 52 45 37
2041
259
240
231
160
155
161
99
85
70
199
187
185
147
143
148
52
44
37 2042 258 238 229 159 154 160 99 84 69
198 186 184 147 142 148 52 44 36
2043
257
236
228
158
153
160
99
83
68
197
185
183
146
141
147
52
44
35 2044 256 235 227 158 153 160 98 83 67
196 184 182 145 140 147 51 43 35
2045
255
234
225
157
152
160
98
82
65
196
183
182
144
140
147
51
43
34 2046 254 233 224 156 152 160 98 81 64
195 182 181 144 140 147 51 42 34
2047
254
232
223
156
151
160
98
80
63
195
181
180
144
139
147
51
42
33 2048 253 231 222 156 151 160 97 80 62
194 181 180 143 139 147 51 42 32
2049
253
230
221
155
151
160
97
79
61
194
180
180
143
139
148
51
41
32 2050 252 229 220 155 151 160 97 78 60
193 180 179 143 139 148 51 41 31
2051
252
228
220
155
151
161
97
78
59
193
179
179
143
139
148
51
41
31 2052 251 227 219 155 150 161 97 77 58 193 179 178 142 139 148 50 40 30
2053
251
227
218
155
150
161
96
76
57
193
178
178
142
138
148
50
40
30 2054 250 226 217 154 150 161 96 75 56 192 178 178 142 138 148 50 39 29
2055
252
228
220
154
150
161
98
78
58
193
179
179
142
138
149
51
41
30
GenCost 2025-26 | 75
Apx Table B.6 Twelve- and twenty-four hour battery cost data by storage duration, component and total costs (multiply by duration to convert to $/kW)
Battery storage (12 hrs)
Battery storage (24 hrs)
Total
Battery
BOP
Total
Battery
BOP
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
Current policies
Global NZE post 2050
Global NZE by 2050
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
$/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh $/kWh
2025
287
287
287
245
245
245
42
42
42
266
266
266
245
245
245
21
21
21 2026 276 273 271 236 234 232 40 39 38
256 254 251 236 234 232 20 20 19
2027
261
260
258
224
223
223
38
37
35
243
242
241
224
223
223
19
18
18 2028 250 248 247 213 214 214 36 34 33
232 231 231 213 214 214 18 17 16
2029
243
240
237
208
207
207
35
33
30
225
224
222
208
207
207
18
16
15 2030 237 233 230 201 200 201 35 33 30
219 216 215 201 200 201 18 16 15
2031
230
225
223
195
193
194
35
32
29
212
209
209
195
193
194
18
16
15 2032 224 218 216 188 186 188 35 32 29
206 202 202 188 186 188 18 16 14
2033
218
211
210
182
180
182
35
32
28
200
195
196
182
180
182
18
16
14 2034 212 205 203 177 173 176 35 31 28
194 189 189 177 173 176 18 16 14
2035
206
198
197
171
167
170
35
31
27
188
183
183
171
167
170
18
16
14 2036 200 192 191 165 161 164 35 31 27
183 177 178 165 161 164 17 15 13
2037
195
186
185
160
156
159
35
31
26
178
171
172
160
156
159
17
15
13 2038 190 180 180 155 150 154 35 30 26
172 165 167 155 150 154 17 15 13
2039
185
175
174
150
145
149
35
30
25
167
160
161
150
145
149
17
15
13 2040 183 173 173 149 144 148 35 30 25
166 159 161 149 144 148 17 15 12
2041
182
172
172
147
143
148
35
30
25
165
157
160
147
143
148
17
15
12 2042 181 171 172 147 142 148 34 29 24
164 156 160 147 142 148 17 15 12
2043
180
170
171
146
141
147
34
29
24
163
156
159
146
141
147
17
15
12 2044 179 169 171 145 140 147 34 29 23
162 155 159 145 140 147 17 14 12
2045
179
168
170
144
140
147
34
29
23
162
154
159
144
140
147
17
14
11 2046 178 168 170 144 140 147 34 28 22
161 154 159 144 140 147 17 14 11
2047
178
167
169
144
139
147
34
28
22
161
153
158
144
139
147
17
14
11 2048 177 167 169 143 139 147 34 28 22
160 153 158 143 139 147 17 14 11
2049
177
166
169
143
139
148
34
28
21
160
153
158
143
139
148
17
14
11 2050 177 166 169 143 139 148 34 27 21
160 152 158 143 139 148 17 14 10
2051
176
166
168
143
139
148
34
27
21
159
152
158
143
139
148
17
14
10 2052 176 165 168 142 139 148 34 27 20 159 152 158 142 139 148 17 13 10
2053
176
165
168
142
138
148
34
27
20
159
152
158
142
138
148
17
13
10 2054 176 165 168 142 138 148 33 26 19 159 152 158 142 138 148 17 13 10
2055
176
165
169
142
138
149
34
27
20
159
152
159
142
138
149
17
14
10
76 | CSIRO Australia’s National Science Agency
Apx Table B.7 Pumped hydro storage cost data by duration, by scenario, total cost basis
$/kW
$/kWh
Current policies Global NZE post 2050 Global NZE by 2050
Current policies Global NZE post 2050
Global NZE by 2050
10hrs
24hrs
48hrs
160hrs
10hrs
24hrs
48hrs
160hrs
10hrs
24hrs
48hrs
160hrs
10hrs
24hrs
48hrs
160hrs
10hrs
24hrs
48hrs
160hrs
10hrs
24hrs
48hrs
160hrs 2025 3960 5687 7394 10601 3960 5687 7394 10601 3960 5687 7394 10601
396 237 154 66 396 237 154 66 396 237 154 66
2026
3884
5578
7253
10399
3884
5578
7253
10399
3884
5578
7253
10399
388
232
151
65
388
232
151
65
388
232
151
65 2027 3810 5472 7115 10201 3810 5472 7115 10201 3810 5472 7115 10201
381 228 148 64 381 228 148 64 381 228 148 64
2028
3736
5366
6977
10003
3736
5366
6977
10003
3736
5366
6977
10003
374
224
145
63
374
224
145
63
374
224
145
63 2029 3659 5255 6833 9796 3659 5255 6833 9796 3659 5255 6833 9796
366 219 142 61 366 219 142 61 366 219 142 61
2030
3582
5144
6688
9589
3582
5144
6688
9589
3582
5144
6688
8514
358
214
139
60
358
214
139
60
358
214
139
53 2031 3601 5171 6724 9640 3604 5176 6730 9649 3616 5193 6730 8595
360 215 140 60 360 216 140 60 362 216 140 54
2032
3615
5192
6751
9679
3621
5201
6762
9696
3642
5231
6762
8657
362
216
141
60
362
217
141
61
364
218
141
54 2033 3631 5215 6780 9721 3640 5227 6797 9744 3668 5269 6797 8720
363 217 141 61 364 218 142 61 367 220 142 55
2034
3646
5236
6808
9761
3658
5254
6831
9794
3697
5310
6831
8789
365
218
142
61
366
219
142
61
370
221
142
55 2035 3661 5258 6836 9802 3677 5281 6866 9845 3727 5352 6866 8859
366 219 142 61 368 220 143 62 373 223 143 55
2036
3676
5280
6865
9842
3696
5308
6902
9896
3756
5395
6902
8929
368
220
143
62
370
221
144
62
376
225
144
56 2037 3692 5302 6894 9884 3715 5336 6938 9947 3786 5438 6938 9001
369 221 144 62 372 222 145 62 379 227 145 56
2038
3707
5325
6923
9926
3735
5364
6974
9999
3817
5482
6974
9073
371
222
144
62
373
224
145
62
382
228
145
57 2039 3723 5347 6952 9968 3755 5392 7011 10052 3848 5526 7011 9147
372 223 145 62 375 225 146 63 385 230 146 57
2040
3739
5370
6982
10011
3774
5421
7048
10105
3879
5571
7048
9221
374
224
145
63
377
226
147
63
388
232
147
58 2041 3750 5386 7003 10041 3790 5443 7077 10146 3906 5609 7077 9284
375 224 146 63 379 227 147 63 391 234 147 58
2042
3762
5403
7024
10071
3805
5465
7105
10187
3933
5648
7105
9348
376
225
146
63
380
228
148
64
393
235
148
58 2043 3773 5419 7045 10101 3820 5487 7134 10228 3960 5687 7134 9413
377 226 147 63 382 229 149 64 396 237 149 59
2044
3784
5435
7067
10132
3836
5509
7163
10269
3987
5726
7163
9478
378
226
147
63
384
230
149
64
399
239
149
59 2045 3796 5452 7088 10163 3851 5531 7192 10311 4015 5766 7192 9543
380 227 148 64 385 230 150 64 401 240 150 60
2046
3807
5468
7110
10193
3867
5554
7221
10353
4043
5806
7221
9610
381
228
148
64
387
231
150
65
404
242
150
60 2047 3819 5485 7131 10224 3883 5576 7250 10395 4071 5846 7250 9677
382 229 149 64 388 232 151 65 407 244 151 60
2048
3831
5502
7153
10256
3898
5599
7280
10437
4099
5887
7280
9744
383
229
149
64
390
233
152
65
410
245
152
61 2049 3842 5518 7175 10287 3914 5622 7309 10480 4128 5928 7309 9812
384 230 149 64 391 234 152 65 413 247 152 61
2050
3854
5535
7197
10318
3930
5645
7339
10523
4157
5970
7339
9881
385
231
150
64
393
235
153
66
416
249
153
62 2051 3864 5549 7215 10344 3944 5665 7365 10560 4184 6009 7365 9945
386 231 150 65 394 236 153 66 418 250 153 62
2052
3873
5563
7233
10370
3958
5685
7391
10597
4211
6047
7391
10009
387
232
151
65
396
237
154
66
421
252
154
63 2053 3883 5577 7251 10396 3972 5705 7417 10635 4238 6087 7417 10074
388 232 151 65 397 238 155 66 424 254 155 63
2054
3893
5591
7269
10422
3986
5725
7444
10672
4266
6126
7444
10140
389
233
151
65
399
239
155
67
427
255
155
63 2055 3902 5605 7287 10448 4000 5745 7470 10710 4293 6166 7470 10206
390 234 152 65 400 239 156 67 429 257 156 64
GenCost 2025-26 | 77
Apx Table B.8 Historical storage cost data, total cost basis
$/kW $/kWh
Battery (1hr) Battery (2hrs) Battery (4hrs) Battery (8hrs) PHES (6hrs) PHES (8hrs) PHES (10hrs) PHES (12hrs) PHES (24hrs) PHES (48hrs) PHES 160hrs) Battery (1hr) Battery (2hrs) Battery (4hrs) Battery (8hrs) PHES (6hrs) PHES (8hrs) PHES (10hrs) PHES (12hrs) PHES (24hrs) PHES (48hrs) PHES 160hrs)
2019
1535
2424
4404
2283
2461
2614
3875
4341
768
606
551
308
249
177
131
88
2020 977 1310 2086 3676 2877 3004
3292 4232 6358
977 655 521 460 387 303
222 142 128
2021
942
1257
1942
3407
2771
3006
3234
4219
6337
942
629
485
426
387
315
226
147
133
2022 1046 1513 2454 4360 3245 3526
3786 4949 7434
1046 756 614 545 481 392
281 184 165
2023
1069
1549
2510
4400
4037
4387
4617
6156
7226
1069
775
627
550
635
517
363
242
170
2024 929 1241 1727 2809
7837
6632 7985
929 621 432 351
768
271 196
2025
778
1050
1540
2464
3960
5687
7394
10601
778
525
385
308
396
237
185
66
Notes: Batteries are large scale. Small scale batteries of 10kWh for home use with 2-hour duration are estimated at $1100/kWh before subsidies (GHD, 2025). However, lower costs may be available for larger household battery sizes.
78 | CSIRO Australia’s National Science Agency
Apx Table B.9 Data assumptions for LCOE calculations
Constant Low assumption
High assumption
Economic life Construction time Efficiency O&M fixed O&M variable CO2 storage
Capital Fuel Capacity factor
Capital Fuel Capacity factor
2025
Years
Years
$/kW
$/MWh
$/MWh
$/kW
$/GJ
$/kW
$/GJ
Gas with CCS 25 2.0 44% 46.0 8.0 8.6
6962 12.7 89%
6962 17.9 53%
Gas combined cycle
25
2.0
51%
36.0
5.0
0.0
2497
12.7
89%
2497
17.9
53% Gas open cycle (small) 25 1.5 36% 17.4 16.1 0.0
2886 12.7 20%
2886 17.9 20%
Gas open cycle (large)
25
1.5
33%
26.0
10.0
0.0
2886
12.7
20%
2886
17.9
20% Gas reciprocating 25 1.1 41% 29.4 8.5 0.0
2022 12.7 20%
2022 17.9 20%
Hydrogen reciprocating
25
1.0
32%
33.0
0.0
0.0
2648
41.9
20%
2648
41.9
20% Black coal with CCS 30 2.0 30% 94.8 8.9 14.1
12941 2.9 89%
12941 5.4 53%
Black coal
30
2.0
42%
64.9
4.7
0.0
6946
2.9
89%
6946
5.4
53% Brown coal 30 4.0 32% 69.0 5.3 0.0
10725 0.6 89%
10725 0.7 53%
Nuclear SMR
30
4.4
33%
200
5.3
0.0
28009
1.1
89%
28009
1.3
53% Nuclear large-scale 30 5.8 33% 200 5.3 0.0 10332 1.1 89% 10332 1.3 53%
Solar thermal
30
1.8
100%
124.2
0.0
0.0
6532
0.0
71%
6454
0.0
57% Large scale solar PV 30 0.5 100% 12.0 0.0 0.0 1621 0.0 32% 1621 0.0 19%
Wind onshore
25
1.0
100%
29.0
0.0
0.0
3248
0.0
48%
3248
0.0
29% Wind offshore (fixed) 30 3.0 100% 175.0 0.0 0.0
5433 0.0 52%
5433 0.0 40%
2030
Gas with CCS 25 2.0 44% 46.0 8.0 8.6
6269 9.6 89%
6252 16.8 53%
Gas combined cycle
25
2.0
51%
36.0
5.0
0.0
2426
9.6
89%
2443
16.8
53% Gas open cycle (small) 25 1.5 36% 17.4 16.1 0.0
2624 9.6 20%
2642 16.8 20%
Gas open cycle (large)
25
1.5
33%
26.0
10.0
0.0
1589
9.6
20%
1600
16.8
20% Gas reciprocating 25 1.1 41% 29.4 8.5 0.0
2010 9.6 20%
2039 16.8 20%
Hydrogen reciprocating
25
1.0
32%
33.0
0.0
0.0
2537
35.7
20%
2574
37.7
20% Black coal with CCS 30 2.0 30% 94.8 8.9 14.1
11961 3.2 89%
12196 5.6 53%
Black coal
30
2.0
42%
64.9
4.7
0.0
6162
3.2
89%
6296
5.6
53% Brown coal 30 4.0 32% 69.0 5.3 0.0
9385 0.7 89%
9588 0.7 53%
Nuclear SMR
30
4.4
33%
200.0
5.3
0.0
21273
0.8
89%
20806
1.0
53% Nuclear large-scale 30 5.8 33% 200.0 5.3 0.0 9658 0.8 89% 9858 1.0 53%
Solar thermal
30
1.8
100%
124.2
0.0
0.0
6074
0.0
71%
6320
0.0
57% Large scale solar PV 30 0.5 100% 12.0 0.0 0.0
847 0.0 32%
1378 0.0 19%
Wind onshore
25
1.0
100%
29.0
0.0
0.0
2629
0.0
48%
2650
0.0
29% Wind offshore (fixed) 30 3.0 100% 175.0 0.0 0.0
4133 0.0 53%
5300 0.0 40%
GenCost 2025-26 | 79
2040
Gas with CCS 25 2.0 44% 46.0 8.0 8.6
4654 9.4 89%
4933 16.4 53%
Gas combined cycle
25
2.0
51%
36.0
5.0
0.0
1939
9.4
89%
1994
16.4
53% Gas open cycle (small) 25 1.5 36% 17.4 16.1 0.0
1767 9.4 20%
1811 16.4 20%
Gas open cycle (large)
25
1.5
33%
26.0
10.0
0.0
1070
9.4
20%
1097
16.4
20% Gas reciprocating 25 1.1 41% 29.4 8.5 0.0
2045 9.4 20%
2120 16.4 20%
Hydrogen reciprocating
25
1.0
32%
33.0
0.0
0.0
2579
26.6
20%
2488
31.2
20% Black coal with CCS 30 2.0 30% 94.8 8.9 14.1
11408 3.0 89%
11561 4.7 53%
Black coal
30
2.0
42%
64.9
4.7
0.0
5624
3.0
89%
5941
4.7
53% Brown coal 30 4.0 32% 69.0 5.3 0.0
8472 0.7 89%
8937 0.7 53%
Nuclear SMR
30
4.4
33%
200.0
5.3
0.0
12216
0.5
89%
16682
0.7
53% Nuclear large-scale 30 5.8 33% 200.0 5.3 0.0 9306 0.5 89% 9795 0.7 53%
Solar thermal
30
1.8
100%
124.2
0.0
0.0
5696
0.0
71%
5837
0.0
57% Large scale solar PV 30 0.5 100% 12.0 0.0 0.0
672 0.0 32%
1191 0.0 19%
Wind onshore
25
1.0
100%
29.0
0.0
0.0
2214
0.0
48%
2253
0.0
29% Wind offshore (fixed) 30 3.0 100% 175.0 0.0 0.0
4126 0.0 54%
5315 0.0 40%
2050
Gas with CCS 25 2.0 44% 46.0 8.0 8.6
4782 9.3 89%
5031 17.4 53%
Gas combined cycle
25
2.0
51%
36.0
5.0
0.0
1978
9.3
89%
2069
17.4
53% Gas open cycle (small) 25 1.5 36% 17.4 16.1 0.0
1798 9.3 20%
1874 17.4 20%
Gas open cycle (large)
25
1.5
33%
26.0
10.0
0.0
1089
9.3
20%
1134
17.4
20% Gas reciprocating 25 1.1 41% 29.4 8.5 0.0
2095 9.3 20%
2222 17.4 20%
Hydrogen reciprocating
25
1.0
32%
33.0
0.0
0.0
2703
24.9
20%
2548
29.6
20% Black coal with CCS 30 2.0 30% 94.8 8.9 14.1
11790 3.0 89%
12282 4.4 53%
Black coal
30
2.0
42%
64.9
4.7
0.0
5791
3.0
89%
6326
4.4
53% Brown coal 30 4.0 32% 69.0 5.3 0.0
8756 0.7 89%
9537 0.7 53%
Nuclear SMR
30
4.4
33%
200.0
5.3
0.0
15355
0.5
89%
13007
0.7
53% Nuclear large-scale 30 5.8 33% 200.0 5.3 0.0
9607 0.5 89%
10429 0.7 53%
Solar thermal
30
1.8
100%
124.2
0.0
0.0
5757
0.0
71%
5344
0.0
57% Large scale solar PV 30 0.5 100% 12.0 0.0 0.0
616 0.0 32%
1029 0.0 19%
Wind onshore
25
1.0
100%
29.0
0.0
0.0
2180
0.0
48%
2207
0.0
29% Wind offshore (fixed) 30 3.0 100% 175.0 0.0 0.0
4231 0.0 55%
5342 0.0 40%
Notes: Economic life is the design life or the period of financing. Total operational life, with refurbishment expenses, is not included in the LCOE calculation but is used in electricity system modelling to understand natural retirement dates. Large-scale solar PV is single axis tracking. The real discount rate for all technologies is 7%. The assumption of 100% efficiency for renewable generation is purely for the purpose avoiding division by zero in the standard LCOE formula. This assumption has no practical impact because fuel cost is zero for these technologies.
80 | CSIRO Australia’s National Science Agency
Apx Table B.10 Electricity generation technology LCOE projections data, $/MWh Category Technology 2025
2030
2040
2050
Low High Low High Low High Low High
Peaking 20% load
Gas open cycle (small)
310
362
264
338
216
288
217
302
Gas open cycle (large) 319 375 215 294 184 261 185 275
Gas reciprocating
244
290
216
281
216
282
218
296
H2 reciprocating 628 628 553 578 453 500 441 486
Flexible load, high emission
Black coal
120
203
113
191
105
176
107
182
Brown coal 167 272 149 246 137 232 141 245
Gas
131
192
108
184
101
171
101
180 Flexible load, low emission Black coal with CCS 224 365 215 353 206 329 211 340
Gas with CCS
214
321
180
297
158
265
159
276
Nuclear SMR 433 718 336 546 207 446 250 360
Nuclear large-scale
200
328
187
312
178
307
183
323 Solar thermal 115 143 108 141 102 132 103 123
Variable
Solar photovoltaic
52
88
29
76
24
67
23
59
Wind onshore 78 129 64 107 55 93 55 91
Wind offshore (fixed)
156
203
126
199
123
200
123
200
GenCost 2025-26 | 81
Apx Table B.11 Hydrogen electrolyser cost projections by scenario and technology, $/kW
Current policies Global NZE by 2050 Global NZE post 2050 Alkaline PEM Alkaline PEM Alkaline PEM
2025
2100
2730
2100
2730
2100
2730 2026 2061 2680 2050 2665 2069 2690
2027
2043
2656
2020
2626
2058
2675 2028 2045 2658 2000 2600 2050 2665
2029
2049
2663
1985
2581
2047
2660 2030 2004 2605 1971 2562 2045 2658
2031
2011
2615
1931
2510
2049
2664 2032 2017 2622 1891 2459 2047 2661
2033
2023
2630
1852
2407
2045
2658 2034 2028 2637 1824 2371 2052 2668
2035
2034
2644
1798
2337
2060
2678 2036 2039 2651 1780 2314 2068 2689
2037
2045
2658
1753
2279
2076
2699 2038 2050 2665 1745 2268 2051 2667
2039
2028
2636
1741
2263
2047
2661 2040 2003 2604 1718 2234 2027 2635
2041
1979
2572
1703
2213
2007
2609 2042 1954 2540 1713 2226 1987 2583
2043
1929
2508
1708
2220
1967
2558 2044 1905 2476 1706 2218 1947 2532
2045
1880
2444
1700
2210
1927
2506 2046 1867 2428 1688 2194 1920 2496
2047
1860
2418
1676
2178
1927
2505 2048 1861 2419 1666 2166 1934 2515
2049
1858
2415
1673
2175
1942
2525 2050 1856 2413 1665 2165 1918 2494
2051
1856
2413
1663
2161
1910
2483 2052 1856 2413 1647 2141 1916 2491
2053
1843
2396
1639
2131
1923
2500 2054 1846 2400 1615 2099 1927 2505
2055
1840
2392
1605
2087
1933
2513
82 | CSIRO Australia’s National Science Agency
Data assumptions
C.1 Technologies and learning rates
The technical approach to applying learning rates is explained in Appendix A and involves a specific mathematical formula. The projection approach uses two global and local learning models (GALLM) which contain applications of the learning formula. One model is of the electricity sector (GALLME) and the other of the transport sector (GALLMT). GALLME projects the future cost and installed capacity of 31 different electricity generation and energy storage technologies and now four hydrogen production technologies. Where appropriate, these have been split into their components of which there are 21 (noting that in total 52 items are modelled). Components have been shared between technologies; for example, there are two carbon capture and storage (CCS) components – CCS technology and CCS construction – which are shared among all CCS plant and hydrogen technologies.
Key technologies are listed in Apx Table C.1 and Apx. Table C.2 showing the relationship between generation technologies and their components and the assumed learning rates under the central scenario. Learning is either on a global (G) basis, local (L) to the region, or no learning (-). Up to two learning rates are assigned with LR1 representing the initial learning rate during the early phases of deployment and LR2, a lower learning rate, that occurs during the more mature phase of technology deployment.
Apx Table C.1 Assumed technology learning rates that vary by scenario Technology Scenario Component LR 1 (%) LR 2 (%) LR 3 (%) References Photovoltaics
Current policies
G
20
30
13
(IEA 2021, IRENA, 2023, Fraunhofer ISE, 2015) Rooftop BOP
L
17.5
8.5
4.5 Large scale BOP
L
17.5
17.5
17.5 Photovoltaics
Global NZE by 2050
G
20
30
23 Rooftop BOP
L
17.5
17.5
12.5 Large scale BOP
L
17.5
17.5
17.5 Photovoltaics
Global NZE post 2050
G
20
30
23 Rooftop BOP
L
17.5
8.5
4.5 Large scale BOP
L
17.5
17.5
17.5 Electrolysis
Current policies
G
10
5
5
(Schmidt et al., 2017, IEA 2024b)
L
8
8
8
GenCost 2025-26 | 83
Electrolysis
Global NZE by 2050
G
18
18
9
L
8
8
8 Electrolysis
Global NZE post 2050
G
10
5
5
L
8
8
8 Ocean
Current policies
G
10
5
5
(IEA, 2021)
Global NZE by 2050
G
20
10
10
Global NZE post 2050
G
14
7
7 Fixed offshore wind
Current policies
G
10
5
5
(Samadi, 2018; Zwaan, et al. 2012; Voormolen et al. 2016; IEA, 2021) Fixed offshore wind
Global NZE by 2050
G
20
10
10 Fixed offshore wind
Global NZE post 2050
G
15
8
8 Floating offshore wind
Current policies
G
10
5
5
G
10
5
5 Floating offshore wind
Global NZE by 2050
G
20
10
10
G
20
10
10 Floating offshore wind
Global NZE post 2050
G
15
8
8
G
15
7.5
7.5 Utility scale energy storage – Li-ion
Current policies
G
7.5
7.5
7.5
(Grübler et al., 1999; McDonald and Schrattenholzer, 2001)
L
7.5
7.5
7.5 Utility scale energy storage – Li-ion
Global NZE post 2050
G
10
10
10
L
10
10
10 Utility scale energy storage – Li-ion
Global NZE by 2050
G
15
15
15
L
15
15
15 Onshore wind
Current policies
G
4.3
4.3
4.3
(IEA, 2021; Hayward & Graham, 2013)
L
9.8
4.8
2.8
Global NZE post 2050
G
4.3
4.3
4.3
84 | CSIRO Australia’s National Science Agency
L
11.3
9.8
4.8
Global NZE by 2050
G
4.3
4.3
4.3
L
11.3
11.3
11.3
While solar photovoltaics are implemented with separate learning rates for large scale and rooftop balance of plant (BOP), inverters are not included in the BOP nor given a learning rate. Instead, they are given a constant cost reduction, which is based on historical data.
The potential for local learning means that technology costs are different in different regions in the same time period. This has been of particular note for technology costs in China, which can be substantially lower than other regions. GALLME uses inputs from GHD (2026) to ensure costs represent Australian project costs. For technologies not commonly deployed in Australia, these costs can be higher than other regions. However, the inclusion of local learning assumptions in GALLME means that they can quickly catch up to other regions if deployment occurs. However, they will not always fall to levels seen in China due to differences in production standards for some technologies. That is, to meet Australian standards, the technology product from China would increase in costs and align more with other regions. Regional labour construction and engineering costs also remain a source of differentiation.
Apx Table C.2 Assumed technology learning rates that are the same under all scenarios Technology Component LR 1 (%) LR 2 (%) LR 3 (%) References Coal, supercritical
-
-
-
Coal, ultra-supercritical
G
2
2
2
(IEA, 2008; Neij, 2008) Coal/Gas/Biomass with CCS
G
20
10
5
(EPRI 2010; Rubin et al., 2007)
L
20
10
5
As above + (Grübler et al., 1999; Hayward & Graham, 2013; McDonald and Schrattenholzer, 2001) Gas peaking plant
-
-
-
Gas combined cycle
-
-
-
Nuclear
G
-
-
(IEA, 2008) Nuclear SMR
G
20
10
5
(Grübler et al., 1999; Hayward & Graham, 2013; McDonald and Schrattenholzer, 2001) Diesel/oil-based generation
-
-
-
Reciprocating engines
-
-
-
Hydro and pumped hydro
-
-
-
Biomass
G
5
5
5
(IEA, 2008; Neij, 2008) Concentrating solar thermal (CST)
G
14.6
7
7
(Hayward & Graham, 2013)
L
14.6
7
7
GenCost 2025-26 | 85
CHP
-
-
-
Conventional geothermal
G
8
8
8
(Hayward & Graham, 2013)
L
20
20
20
(Grübler et al., 1999; Hayward & Graham, 2013; McDonald and Schrattenholzer, 2001) Fuel cells
G
20
10
10
(Neij, 2008; Schoots, Kramer, & van der Zwaan, 2010) Steam methane reforming with CCS
G
20
10
5
(EPRI, 2010; Rubin et al., 2007)
L
20
10
5
As above + (Grübler et al., 1999; Hayward & Graham, 2013; McDonald and Schrattenholzer, 2001)
To provide a range of capital cost projections for all technologies, we have varied learning rates for technologies where there is more uncertainty in their learning rate. We focus on variable renewable energy and storage given that these technologies tend to be lower cost and crowd out opportunities for competing low emission technologies. Apx. Figure C.1 shows the learning rates by scenario for solar PV, electrolysis, ocean energy (wave and tidal), offshore wind, batteries and pumped hydro. The remainder of learning rate assumptions, which do not vary by scenario are shown in Apx. Table C.2.
In addition to the offshore wind learning rate, we have included an exogenous increase in the capacity factor up to the year 2050 of 0.9% per year. This assumption extrapolates past global trends (see Appendix D). As discussed in Appendix D, Australia has had a flat onshore wind capacity factor trend and so these global assumptions do not apply to Australia. The capacity factor for floating offshore wind is assumed to be 5.6% higher than that of fixed offshore wind, based on an average of values (Wiser et al., 2021). Capacity factors for offshore wind are assumed to improve in Australia in line with the rest of the world.
C.2 Electricity demand and electrification
Various elements of underlying electricity demand are sourced from the World Energy Outlook (IEA, 2021; IEA, 2022; IEA, 2023; IEA, 2024; IEA, 2025). Demand data is provided for the Stated Policies scenario, which is used in our Global NZE post 2050 scenario. The demand data from the Current Policies scenario is used in our Current policies scenario. Global NZE by 2050 demand is sourced from the Net Zero Emissions by 2050 scenario. We also allow for some divergence from IEA demand data in all scenarios to accommodate differences in our modelling approaches and internal selection of the contribution of electrolysis to hydrogen production.
C.2.1 Global vehicle electrification
Global adoption of electric vehicles (EVs) is projected using an adoption curve calibrated to correspond the matching scenarios from the IEA World Energy Outlook. The shape of the adoption curve varies by vehicle type, where cars and light commercial vehicles (LCV) have faster rates of adoption, followed by medium commercial vehicles (MCV) and buses. The adoption rate is applied to new vehicle sales shares.
86 | CSIRO Australia’s National Science Agency
C.3 Hydrogen
In GenCost projections prior to 2022-23, hydrogen demand was imposed together with the type of production process used to supply hydrogen. In our current model, GALLME determines which process to use – steam methane reforming with or without CCS or electrolysers. This choice of deployment also allows the model to determine changes in capital cost of CCS and in electrolysers.
The model does not distinguish between alkaline (AE) or Proton Exchange Membrane (PEM) electrolysers. That is, we have a single electrolyser technology. The approach reflects the fact that GALLME is not temporally detailed enough to determine preferences between the two technologies which are mainly around their minimum operating load and ramp rate. There is currently a greater installed capacity of AE which has been commercially available since the 1950s, whereas PEM is a more recent technology.
The IEA have included demand for electricity from electrolysis in their scenarios. Since GALLM is endogenously determining which technologies are deployed to meet hydrogen demand, we have subtracted the IEA’s demand for electricity from electrolysis from their overall electricity demand. The assumed hydrogen demand assumptions for the year 2050 are shown in Apx. Table C.3 and include existing demand, the majority of which is currently met by steam methane reforming. The reason for including existing demand is that in order to achieve emissions reductions the existing demand for hydrogen will also need to be replaced with low emissions sources of hydrogen production.
Apx Table C.3 Hydrogen demand assumptions by scenario in 2050 Scenario Total hydrogen demand (Mt) Current policies 128
Global NZE post 2050
130 Global NZE by 2050 342
C.4 Government climate policies
Carbon trading markets exist in major greenhouse gas emitting regions overseas at present and are a favoured approach to global climate policy modelling because they do not introduce any technological bias. We directly impose the IEA carbon prices. The IEA also includes a broad range of additional policies such as renewable energy targets and planned closure of fossil fuel-based generation. The GALLME modelling includes these non-carbon price policies as well but cannot completely match the IEA implementation because of model structural differences. The IEA have greater regional and country granularity and are better able to include individual country emissions reduction policies. Some policies are difficult to recreate in GALLME due to its regional aggregation. Where we cannot match the policy implementation directly, we align our implementation of non-carbon price policies so that we match the emission outcomes in the relevant IEA scenario.
GenCost 2025-26 | 87
We align our scenarios with the IEA and the IEA does not include more recent announcements or changes of government policy since the IEA report was complete. As such, the country policy commitments included are not completely up to date.
C.5 Resource constraints
The availability of suitable sites for renewable energy farms, available rooftop space for rooftop solar PV and sites for storage of CO2 generated from using CCS have been included in GALLME as a constraint on the amount of electricity that can be generated from these technologies (Apx. Table C.4) (see Government of India, 2016, Edmonds, et al., 2013 and Hayward and Graham, 2017 for more information on sources). With the exception of rooftop solar PV these constraints are removed in the Global NZE by 2050. Floating offshore wind has some technical limitations in regions, but these limitations are greater than electricity demand.
C.6 Other data assumptions
GALLME international black coal and gas prices are based on (IEA, 2023) with prices for the Stated Policies scenario applied in all cases. The IEA tends to reduce its fossil fuel price assumptions in scenarios with stronger climate policy action. Whilst we agree that stronger climate policy action will lead to lower demand for fossil fuels, we do not think it follows that fossil fuel prices must fall23. This is one of the very few areas where we do not align with all IEA scenario assumptions. Brown coal is not globally traded and has a flat price of 0.6 $/GJ.
Apx Table C.4 Maximum renewable generation shares in the year 2050, except for offshore wind which is in GW of installed capacity. Region Rooftop PV % Large scale PV % CST % Onshore wind % Fixed offshore wind GW AFR 21 NA NA NA NA
AUS
35
NA
NA
NA
NA CHI 14 NA NA NA 1073
EUE
21
NA
NA
NA
NA EUW 21 2 23 22 NA
FSU
25
NA
NA
NA
NA
23 In the long run, fossil fuel prices will fluctuate due to cycles of demand and supply imbalances. However, underlying these fluctuations, prices should track the cost of production given the competitive nature of commodity markets. This relationship holds whether demand is falling or rising over the long run.
88 | CSIRO Australia’s National Science Agency
Region Rooftop PV % Large scale PV % CST % Onshore wind % Fixed offshore wind GW IND 7 21 18 4 302
JPN
16
1
12
11
10 LAM 25 NA NA NA NA
MEA
21
NA
NA
NA
NA NAM 30 NA NA NA NA
PAO
11
1
8
8
15.5 SEA 14 3 32 8 NA
NA means the resource is greater than projected electricity demand. The regions are Africa (AFR), Australia (AUS), China (CHI), Eastern Europe (EUE), Former Soviet Union (FSU), India (IND), Japan (JPN), Latin America (LAM), Middle East (MEA), North America (NAM), OECD Pacific (PAO), Rest of Asia (SEA), and Western Europe (EUW)
Power plant technology operating and maintenance (O&M) costs, plant efficiencies and fossil fuel emission factors were obtained from (GHD, 2025) (IEA, 2016b) (IEA, 2015), capacity factors from (IRENA, 2023) (IEA, 2015) (CO2CRC, 2015) and historical technology installed capacities from (IEA, 2008) (Gas Turbine World, 2009) (Gas Turbine World, 2010) (Gas Turbine World, 2011) (Gas Turbine World, 2012) (Gas Turbine World, 2013) (UN, 2015a) (UN, 2015b) (Energy Information Administration, 2017a) (Energy Information Administration, 2017b) (GWEC) (IEA, 2016a) (World Nuclear Association, 2017) (Schmidt, Hawkes, Gambhir, & Staffell, 2017) (Cavanagh, et al., 2015).
New capacity that was installed in 2023 was sourced from (IRENA, 2024) (Global Energy Monitor, 2024a) (Global Energy Monitor, 2024b) and (Global Energy Monitor, 2024c).
GenCost 2025-26 | 89
Frequently asked questions
The following list of questions represents a summary of the most commonly asked questions in relation to methods and assumptions applied in GenCost.
D.1 Process
D.1.1 Why does GenCost not immediately change its report when provided with new advice from experts?
The GenCost report undertakes a significant stakeholder consultation process, but it is not a consensus process and the response to feedback is based on its quality, not who provided it. This process is consistent with the objectivity and scientific approach that stakeholders expect of CSIRO.
There have been suggestions from some stakeholders that because some information was provided by an expert or group of experts it should have been accepted and acted upon immediately. This is not sufficient grounds for making a change to the GenCost report. Changes to the GenCost report need to be based on public evidence and reason. They cannot be based on assertions alone, no matter the qualifications and experience of the individual or group of individuals providing input.
GenCost reserves the right to test the quality of any evidence provided. There are widely varying qualities of data and evidence provided in the consultation process. Stakeholders should consider the many issues that can impact the quality of evidence when providing it such as the appropriateness of methodologies used to develop the data, stated or unstated vested interests behind the data development, and the level of inherent proof the evidence represents (e.g., correlation versus causation, opinion versus verifiable data).
Finally, CSIRO reserves the right to prioritise the issues and evidence it chooses to investigate. Not every topic raised will be fully investigated in the year the feedback is received. We prioritise issues based on their relevance, the weight of feedback received, and the technical challenges associated with investigating the topic in a way that meets our own standards.
D.2 Scenarios
D.2.1 Why are disruptive events and bifurcations excluded from the scenarios?
It is acknowledged that the future evolution of major drivers of the global energy system will not be smooth, particularly considering the recent pandemic and Ukraine war impacts on the energy sector. GenCost provides relatively smooth projections of capital costs over time compared to what is likely to occur. This reflects our understanding that very few end-users of the capital cost projections would like to access results that include major discontinuities. More volatility in inputs
90 | CSIRO Australia’s National Science Agency
will lead to more volatility in all model outputs. Such volatility can interfere with the interpretation of models which are often seeking to answer separate questions about the evolution of the system by reading into the changes in the modelling results. As such, our judgment is that adding more realism does not add value in this case.
D.2.2 Why is no sensitivity analysis conducted and presented?
The staff delivering GenCost have many decades of experience in energy and electricity system modelling. They understand which parameters in the model have the greatest impact on model outcomes. The scenarios have been designed to explore those parameters that are the most uncertain and impactful (within a plausible range) so that they provide a set of results that represent the likely range of outcomes. The possible range of outcomes is wider and could be calculated. However, our understanding of end-user needs is that they require outputs that align with globally accepted literature on the likely range of major drivers such as global climate policy, learning rates and resource constraints. Should our understanding of the likely range of any of these factors change, the scenarios will be updated.
D.3 Capital costs
D.3.1 What did you base your large-scale nuclear costs on?
The GenCost 2023-24 final report provides a detailed discussion of the method for estimating large-scale nuclear costs in Section 2.5
D.3.2 Why have the estimates for nuclear SMR capital costs increased so much since 2022?
The GenCost 2023-24 final report provides a detailed discussion of the history of estimating nuclear SMR costs in Section 2.4. This report has adopted the project cost for the Darlington nuclear SMR project as its primary source for current and near term costs.
D.3.3 Do you assume Australia continues to rely on overseas technology suppliers or are you assuming Australia develops its own original equipment manufacturing capability?
The context of this question is the concern that reliance on overseas manufacturers makes Australia vulnerable to non-competitive market pricing (e.g., the dominance of China), delayed access to technology because of competing buyers or represents a security of supply risk in the event of conflict in or with supplying countries. In this context, some government policies have provided international partnership support and direct grants for critical minerals projects24.
24 https://www.industry.gov.au/publications/critical-minerals-strategy-2023-2030/our-focus-areas/2-attracting-investment-and-building-international-partnerships
GenCost 2025-26 | 91
Whilst GenCost will continue to monitor these developments, the equipment component of capital cost estimates remains based on the best available representative technology cost deployment in Australia with equipment supplied from anywhere in the world that meets our standards.
D.3.4 Why does GenCost persist with the view that technology costs will fall over time when there are many factors that will keep technology costs high?
In the GenCost 2022-23 final report, research was outlined that indicated that there is no historical precedence for the real cost of commodities increasing indefinitely in real terms. Most periods of high prices resolve themselves within 4 years. Longer-term commodity price super cycles do occur but are shallower and are associated with changes in global economic growth. There is no suggestion from stakeholders that the world is in a major economic growth cycle. It was also argued in GenCost 2022-23 that global manufacturing will not need to be endlessly scaled up. Rather global technology capacity forecasts indicate that technology manufacturing capacity will need to grow to 2030, but after that point will be able to meet mostly linear demand for additional capacity without significant additional scale-up.
Stakeholders have raised the following additional points on this topic:
•
That the energy sector may have a different inflationary path to the economy in general
•
That GenCost needs to prove that the world is not in a new commodity super cycle
•
That concentration of manufacturing in China will lead to non-competitive behaviour and high prices for those products, particularly solar
•
That demand for energy technologies will remain non-linear for a long time because of delays in Australian deployment.
The current uncertainty in global manufacturing is acknowledged and makes forecasting at this time in history very challenging. The global inflationary event triggered by the pandemic is a significant structural break. Based on the evidence available of similar events, the approach taken has been to assume a reasonably quicker resolution of high technology prices with some lingering effects for 3 to 6 years, the length depending on the scenario. Recovery has started but has also been delayed by the 2026 Iran war.
The data on technology project costs from GHD and various commodities price inputs to those technologies indicates (so far) that the evidence is in alignment with our approach. Some costs have already fallen in real terms. Some are still rising but the rate of increase is significantly lower. The evidence from GHD (2026) points to cost pressures easing. Commodity price reporting also indicates cost pressures have eased in raw material markets such as lithium, not withstanding recent volatility associated with the Iran war.
Based on this data, it does not appear energy is on a different path to the rest of the economy. Solar panels produced predominantly by China are recovering better than others and their price increase was more modest to begin with.
Regarding the expected linear growth rates in technology deployment, this refers to the global technology deployment and the required global manufacturing capacity to meet this growth.
92 | CSIRO Australia’s National Science Agency
Australia’s technology deployment rate, while important to us as Australians, has very little impact on the scale or cost of global technology manufacturing.
Notwithstanding these points, our projection methodology assumes increasing land and installation costs (in real terms). These exceptions are due to the scarcity of land and suitably qualified construction labour. This assumption means that the costs of some technologies (particularly mature technologies) increases for significant parts of the projection period.
D.3.5 Why is the uncertainty in the data not emphasised more?
GHD (2026) provide an uncertainty range of +/- 30% for their capital costs. To reduce this uncertainty, their analysis would have to be performed on a specific project. The GenCost project requires general data, not specific project data, that can be used in national level modelling studies. GHD (2026) also provide factors to convert the general costs to specific locations in the National Electricity Market (NEM). In that context, GenCost data can be converted to specific locations. An important aspect for GenCost is that all data is on a common basis.
Some stakeholders have requested that we emphasise this uncertainty in capital costs more in the text and diagrams. The main purpose of GenCost has always been to provide data which can be used in modelling studies. While there are stochastic modelling frameworks, the majority of electricity system models used in Australia are deterministic. In simple terms, this means they use single data points without any probability information attached to them. Therefore, GenCost capital cost outputs, which focus on providing scenarios to explore uncertainty rather than probability ranges, remain appropriate for the end-use they are created for.
LCOE data is specifically designed for the non-modelling community. In this case, we take a different approach. LCOE data is always presented as a range representing the plausible maximum and minimum costs. We also provide ranges for key inputs to the LCOE calculations such as capital costs, fuel costs and capacity factors.
D.3.6 Why include an advanced ultra-supercritical pulverised coal instead of cheaper, less efficient plant designs?
Some stakeholders take a view that although Australia has national and state net zero emissions policies by 2050, the highest greenhouse gas emitting options should remain on the table. The deployment of new coal has low plausibility given its high emissions intensity. A high efficiency design brings it closer to being plausible by reducing its emissions. Perhaps the most plausible scenario for building new coal consistent with meeting the net zero emissions by 2050 target would be to later retrofit coal generation with carbon capture and storage. Carbon capture and storage imposes a very significant fuel efficiency loss on the coal generator. In this context, it is even more important to start from a high efficiency coal generation technology.
GenCost 2025-26 | 93
D.4 LCOE
D.4.1 Why is the economic life used in LCOE calculations instead of the full operational life?
The LCOE calculation converts all upfront and ongoing costs to annual costs which is then divided by annual production. The capital cost component of a technology is converted to an annual repayment to the debt and equity providers. The annual repayment amount is determined using the economic life and the weighted average cost of capital. The economic life is shorter than the asset life for some technologies such as coal, nuclear and hydro. Some stakeholders have queried why this is so.
Debt and equity providers require a shorter payback period than the total asset life for some technologies to avoid the risk that part of the equipment might fail or might need new investment (sometimes called refurbishment or extension costs) to keep operating safely and reliably. To determine the economic life, debt and equity providers might look to the warranties provided with the equipment. They might also look at the typical timing of refurbishments or life extensions for that technology. The economic life is an input provided by the engineering firm that AEMO commissions each year as an input to GenCost.
Some stakeholders suggested that coal and nuclear could access special financing arrangements to move the economic life closer to the asset life. However, our preference is not to introduce special arrangements for technologies where there is limited Australian evidence. A common approach to the LCOE calculation is important to maintain comparability. The 2024-25 report does explore the impact of longer capital recovery periods in Section 2. It finds there is no significant benefit from the longer operational life of nuclear relative to shorter-lived technologies whose costs have been falling over time.
Determining the economic life of storage is more complex because the cycle life comes into play in determining the life of some components. The cycle life and intended use of the storage device might also be something debt and equity providers are also interested in to set the repayment date. Batteries in GenCost are costed for a project which has purchased a 20-year warranty on the battery (this warranty is costed as part of the ongoing operating and maintenance cost – see GHD (2026) for more information on this).
It should also be noted that cycle life is often calculated in the academic literature based on a full charge and discharge and is tested over a shorter period than would occur in practice. It is not clear how well deployed storage projects will match the lab tests. Their operation may be more prone to partial discharge, preferring to save some charge for higher priced periods. That is, they will bid parts of their storage capacity at different prices. Time will tell how this bidding behaviour will impact their cycle life, but it is a reasonable expectation that practical operation will be less damaging to batteries than the lab tests.
94 | CSIRO Australia’s National Science Agency
D.4.2 Coal and nuclear plants are capable of very high capacity factors, why do LCOE calculations not always reflect this?
Stakeholders are sometimes not aware of the difference between the availability factor, which is how often a plant will be technically available to generate electricity and the capacity factor which is how often they typically generate electricity after the effects of competition or other market constraints which limit generation.
In the last ten years in Australia, baseload generators have had an average capacity factor of 59% (see Appendix D GenCost 2022-23 final report). The simple reason for this outcome is that most baseload plants need to reduce production at night and in milder seasons when demand is lowest. There are individual generators that do achieve around 90%. These are a minority of plants which have a fuel cost advantage which allows them to keep running at full production during low demand periods by underbidding other generators for the right to keep generating at a high level.
GenCost LCOE calculations allow for the fact that a new baseload generator might achieve a capacity factor of up to 89% based on the maximum achieved by black and brown coal generators. At the low end of the range a capacity factor of 53% is assumed for new black coal, brown coal or nuclear generators which is equivalent to achieving 10% below the average capacity factor for black and brown coal. Around 10% of nuclear generators globally run at less than 60% capacity factor and many have run at over 90%25. However, we prefer to use Australian data for the plausible baseload plant operation data because it is consistent with our electricity load curve while other countries may have very different loads. For example, some equatorial and northern regions with hotter and colder climates have higher rates of air conditioning in buildings leading to flatter electricity loads (where either electricity or combined heat and power are the energy source).
Higher penetration of renewables, which have a zero fuel cost, could make it difficult for new baseload plant to achieve high capacity factors depending on the scale of demand overall. Ultimately, we do not know what new coal or nuclear will be competing with in the future. The key principle though is to acknowledge a plausible range rather than assume only the best outcome for new build capacity factors.
D.4.3 Why do LCOE calculations not use the lowest historical capacity factors for the low range assumptions?
For all existing technologies there are some generators that are performing poorly relative to what might be expected, and these represent the low range of historical capacity factors which were examined in Appendix D of the GenCost 2022-23 report. The data does not reveal why some projects are performing below expectations, but it could represent older technologies or, for renewables, sites that did not live up to expectations in terms of the renewable resource. GenCost LCOE capacity factor low range assumptions are developed on the basis that new entrant technologies will not be deployed if they cannot perform close to the current average capacity factor performance. Investors would prefer to avoid such projects in preference for more
25 https://world-nuclear.org/our-association/publications/world-nuclear-performance-report/global-nuclear-industry-performance
GenCost 2025-26 | 95
attractive investment options. Accordingly, we apply a common rule across renewables, coal, nuclear and gas that the minimum capacity factor for new plant is 10% below the ten years average capacity factor for that technology or its nearest equivalent grouping (baseload technologies are treated as one group).
D.4.4 Why were all potential cost factors not included in the LCOE calculations?
While each technology has its own specific characteristics the goal of the LCOE calculation is to use a common formula to calculate costs so that that observed differences in costs are due to a small set of key differences in the technology, namely: capital costs, fuel costs, fuel efficiency, operating and maintenance costs, economic life and construction time. However, often stakeholders request that other special topics be included in the calculations. Items requested to be added to the LCOE analysis by stakeholders include:
•
Plant decommissioning and recycling costs
•
Deeper pre-development costs
•
Technology degradation
•
Whole-of-life emissions
•
Savings from developing on a brownfield site
•
Various environmental impacts
•
Energy in manufacturing costs
•
Public acceptance barriers
•
National security impacts
•
Extreme climate events
•
Connection costs
•
Marginal loss factors.
Adding these additional parameters would greatly expand the physical and time boundary of the generic generation projects assumed in GenCost and require more complicated formulas to implement. Our current understanding is that few of the topics presented in the feedback have a large enough impact on LCOE to warrant a change in the boundary or formula. That is, it would add complexity and cost to the project without significantly changing the outcome of the comparisons. Some factors, like marginal loss factors are significant but are too unpredictable at this stage of the energy transition.
We do acknowledge that taking account of brownfield project characteristics would make a difference in costs. This is because brownfield projects can avoid some development costs associated with site selection, grid connection and land. However, brownfield projects are outside our stated scope for GenCost of greenfield or new build projects. The study of brownfield projects is always site-specific and more resource intensive and for these reasons less generally comparable to other options. Their inclusion would essentially amount to bringing “one-off” projects into the analysis. This is inconsistent with our goal of providing a general comparison metric. Some brownfield project costs are included in AEMO’s publicly accessible forecasting input data.
96 | CSIRO Australia’s National Science Agency
There are two exceptions in the past where GenCost added new technology cost elements. These are CO2 storage costs for carbon capture and storage technologies and integration costs for variable renewables. In both cases, the impact of these additional elements is significant and justifies modification of the standard approach to LCOE calculation.
Given that GenCost does not account for all potential additional project costs such as those captured in the list above, real projects are likely to cost more than indicated by the LCOE. Consequently, investors must do their own deeper studies to discover these. Likewise, investors who are interested in brownfield project development will need to source this information elsewhere (e.g., check AEMO publications) or do their own analysis.
Energy used in manufacturing costs are accounted for in capital costs. Notwithstanding the current difficulties in manufacturer profitability following the global supply chain crunch, to remain solvent, manufacturers must recover these costs (as with all other costs), in the long term, by building them into their technology prices. Also, the more that global economies track and potentially price greenhouse gas emissions, the greater the incidence of lifecycle greenhouse gas emissions of projects being built into technology prices. Planned carbon border adjustment mechanisms are an example of this.
D.4.5 What is the boundary of development costs? Is it only costs from the point of contracting a developer before commencing construction?
GHD’s reports and data break down the capital cost into three components: equipment, land and development and installation costs. Development costs are captured in the land and development segment. GHD (2026) provides this definition of the land and development cost component:
“The development and land costs for a generation or storage project typically include the following components:
•
Legal and technical advisory costs
•
Financing and insurance (no interest during construction considered)
•
Project administration, grid connection studies, and agreements
•
Permits and licences, approvals (development, environmental, etc)
•
Land procurement and applications.”
D.4.6 How is interest lost during construction included in GenCost?
The type of capital cost data included in GenCost is called overnight capital costs. That is, it is the cost if you built it overnight. Consequently, to make the costs more realistic, interest lost during the construction period needs to be added when using this data.
Interest lost during construction is added differently depending on how the data is being used. When overnight capital cost data is being used in an energy system model, information is provided to the model about the construction time. The time discounting function within the system model accounts for the interest lost during construction in the time delay between investment expenditure and when the project is fully operational.
GenCost 2025-26 | 97
When overnight capital cost data is being used in an LCOE calculation a different approach is used. LCOE calculations must average all costs into a single year of electricity production and so the time during construction does not exist as a concept. However, there are several ways in which the interest lost can be added to an LCOE. GenCost uses the simplest way which is to increase the capital cost by the assumed discount rate raised to the power of the construction time26. There are more sophisticated ways to do this which account for developer plans for drawing down the financing during construction depending on the arrival time of different plant parts and payment for each component. These more detailed approaches are appropriate for real project planning but require tailored calculations for each technology and a cash flow model approach. The cashflow approach tracks payments over each year of construction plus economic life before averaging them into a single yearly cost (dividing total expenditure including the construction period by total production including periods of zero production during the construction period). The simpler approach is more efficient (requires just a few cells of calculations and fewer input data), but the latter is more accurate. The simpler approach tends to overestimate interest lost during construction as it assumes all funds need to be drawn down at the beginning of construction.
D.4.7 Why are the cost of government renewable subsidies not included in the LCOE calculations for variable renewables with integration costs?
The cost of government subsidies for variable renewables, in whatever form they take, are not included as a cost because all of the variable renewable costs applied in the modelling are without subsidy. In other words, because we do not subtract any subsidies from the cost of variable renewable generation, it is not necessary to add those subsidies back in as a cost to society. The GenCost estimates of the cost of integrating variable renewables are without any government subsidies.
D.4.8 Why is a value of 100% applied to the fuel efficiency of renewables in the LCOE formula?
For our purposes there is no practical limit to supply of solar and wind power and its cost as a fuel is free. Since the fuel price applied is zero, any value for renewable energy efficiency other than zero would work in the fuel cost formula (and avoid division by zero) where fuel cost equals FuelPrice÷FuelEfficiency. We choose 1 or 100% for simplicity. This is not to say that the energy conversion efficiency of renewable generation technologies is 100%, or irrelevant, or not accounted for. The conversion efficiency of solar irradiance and wind to electricity is accounted for in the capital cost. Manufacturers apply a nameplate plant capacity in watts to the equipment they sell based on exposure to representative wind speeds or solar irradiance and this reflects the energy conversion efficiency of the plant. Conversion efficiency is also partially captured in land costs which reflect the scarcity of sites with the required renewable resources to operate at nameplate capacity.
26 GenCost readers who have downloaded the Appendix tables from CSIRO’s Data Access Portal should be able to find this step in the cell formula under the Capital component of the LCOE calculation
98 | CSIRO Australia’s National Science Agency
D.4.9 Why do you apply only one discount rate or weighted average cost of capital to all technologies?
This question may arise in the context of stakeholder concerns that some projects might be government funded and receive a lower financing rate and that should be included. While GenCost recognises that governments have in the past and may choose in the future to provide lower cost financing to selected projects, GenCost makes no specific assumptions about who will invest in a technology project.
Another factor guiding our approach is that we wish to compare technologies on a common basis wherever that approach does not lead to an unwanted distortion. In most cases, that can be achieved but there are exceptions. In some cases, we need to apply a different formula or method to different technologies to capture important additional costs such as adding reliability costs for variable renewables or carbon dioxide storage costs for CCS technologies (see D.4.4 for a longer discussion of what additional costs we have chosen to include).
Previous versions of GenCost also applied a cost of capital premium to fossil fuel technologies due to their additional climate policy risk. However, our judgement was that although that risk is real and ongoing, we were no longer able to find a cost of capital premium that adequately captured that risk. Instead, wherever we present high emission fossil fuel technology costs we simply state that investment in these technologies may not be consistent with government emission targets.
In conclusion, our judgment is that, in the case of the cost of capital, applying the same rate to every technology is the most informative and least distortionary approach for levelised cost of electricity. Other modelling exercises may take an alternative approach. However, our LCOE data is not likely to be an input to any detailed electricity system modelling. Rather LCOE data is simply an indicator of the potential direction of the results from more detailed modelling.
D.4.10 Why did you take the maximum and average of existing generator prices to create the high and low range new build coal prices?
Our goal is to explore the high and low range for total coal generation costs in the LCOE calculations. To do this we include high and low ranges for the various inputs to coal generation costs such as capacity factors, capital costs and coal fuel costs.
We require coal prices for new-build projects which are different to coal prices that are received by existing generation sites. Some existing generators receive low coal prices because they may have captured an adjacent coal mine with no competing rail line to export markets. Alternatively, if they are competing with export markets, they are more likely to have developed a favourable long-term contract to manage high price risk. New-build projects will start their life by competing with export markets for supply of coal.
High and low coal prices are sourced from the AEMO Inputs and Assumptions workbook. The June 2022 Inputs and assumptions workbook provided coal prices for new build and existing coal generators. Reflecting the issues discussed above, average new build coal prices were two and half times higher than the minimum existing generator coal prices. For GenCost 2022-23, our methodology for selecting coal prices to use in GenCost was to take the minimum and maximum of only the new build coal prices.
GenCost 2025-26 | 99
After June 2022, AEMO has no longer published new build coal prices. AEMO continued to publish coal prices, but only for existing generators which remain in the system.
To create the high and low range for new build coal prices post-2022-23 GenCost had to apply a new methodology based on the only available data which was coal prices for existing generators. Knowing that new build coal prices are at least as high as that for existing generators, for the maximum, GenCost simply takes the maximum of existing generator prices.
However, for the minimum new build coal prices, taking the minimum of existing generator prices is not appropriate. CSIRO developed a new methodology, using the only available data from AEMO on coal prices for existing generators, to extrapolate the low-cost range. This methodology takes into account that new-build coal generation projects cannot achieve the same low prices as existing generators, hence why the low coal prices are averaged. The average of the lowest coal price trajectory for existing generators tends to be two to three times the minimum coal price for those generators, which maintains the previously observed relationship between existing generator and new build coal prices.
IEA coal prices are used in the global modelling which underpins the capital cost projections. A different source is justified on the basis that the global modelling requires a consistent set of global fuel prices by major global region which is not available from AEMO which only provides Australian data.
D.4.11 Why do you not include high and low ranges for economic life?
Economic life is in some cases set by a warranty. This is the case for batteries. In other cases, it represents long standing practice in the financing of utility assets which are unlikely to vary significantly between Australian projects. While many stakeholders have provided evidence for variation in asset lives, there has been little evidence provided on variation in economic life or warranties or loan periods. At this stage, there is not enough information to form a basis for a high and low range for economic life as an input to the LCOE calculations. See D.4.1 for a discussion on the differences between economic and asset life.
D.4.12 Why are your low range capacity factors for coal and renewables closer to the historical average capacity factor?
In the GenCost 2022-23, report capacity factors from the previous ten years were reviewed to inform our choices about capacity factors in the LCOE calculations. Stakeholders have noted that the low range capacity factor applied is close to the ten-year average capacity factor. In fact, the approach to set the low range value for new-build generators is to use a value 10% below the average capacity. Our reasoning is that new projects are less likely to proceed if their capacity factor is significantly lower than the market average. The same method is applied for renewables as for coal to develop the low range capacity factor assumption.
For the high capacity factor assumption, the highest capacity factor achieved over a ten year period is applied. Given these are new-build, it is appropriate to be less conservative on the high range assumption. Again, the approach is the same for coal and renewables.
100 | CSIRO Australia’s National Science Agency
D.4.13 If GenCost shows renewables are cheaper, why are electricity prices higher in Australia and in countries transitioning to renewables?
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 2022, global natural gas supply constraints, triggered by sanctions on Russia due to the Ukraine war, together with unplanned coal plant outages caused a price spike in Australia that is still reverberating through the electricity system. The prices of other electricity systems around the world were also impacted by the rising global fossil fuel prices and constrained supply of gas.
In Australia, retailers, experiencing these conditions, secured electricity supply contracts for the future which factored these higher prices in. A decrease in gas prices or growth in new supply capacity (net of retirements) can put downward pressure on market prices. However, there is no guarantee that either of these forces will maintain downward pressure on prices. If gas prices rise again or capacity is retired faster than it is rebuilt, then prices will increase again regardless of the cost of new entrant capacity.
The quality of both renewables and fossil fuel resources varies substantially around the world as do the pace of transition to lower emission sources, the degree of state ownership, subsidies, age of generation fleet and market incentives for building new capacity. As a result, due to the variety of differences in circumstances and the impact of supply and demand imbalances, there are no clear causal relationships that can be concluded from a simple correlation analysis of electricity prices and the energy source used by country or region.
D.4.14 If nuclear has such high capital costs why do they have such low-cost nuclear electricity overseas?
New large-scale nuclear costs are significantly lower than nuclear SMR but both represent moderate- to high-cost sources of electricity generation. This result could be perceived as out of step with overseas experience where some countries enjoy low-cost nuclear electricity. There are two reasons for this seemingly inconsistent result.
The first is that new generation technology electricity costs have only weak transferability between countries. While the technology can be identical, electricity generation costs vary widely between countries due to differences in installation, maintenance and fuel costs in each country. There are also unknown or known subsidies and different levels of state versus private ownership which impact the costs that ultimately get passed to electricity customers.
The second issue is that observations of low-cost nuclear electricity overseas are in most cases referring to historical rather than new projects which could have been funded by governments or whose capital costs have already been recovered by investors. Either of these circumstances could mean that those existing nuclear plants are charging lower than the electricity price that would be
GenCost 2025-26 | 101
required to recover the costs of new commercial nuclear deployment. Such prices are not available to countries that do not have existing nuclear generation such as Australia.
In summary, given overseas new generation electricity costs are not easily transferable and may be referring to assets that are not seeking to recover costs equivalent to a commercial new-build nuclear plant, there may be no meaningful comparison that can be made between overseas nuclear electricity prices and the costs that Australia could be presented with in building new nuclear.
102 | CSIRO Australia’s National Science Agency
Technology inclusion principles
GenCost is not designed to be a comprehensive source of technology information. To manage the cost and timeliness of the project, we reserve the right to target our efforts on only those technologies we expect to be material, or that are otherwise informative. However, the range of potential futures is broad and as a result there is uncertainty about what technologies we need to include.
The following principles have been established to provide the project with more guidance on considerations for including technology options.
E.1 Relevant to generation sector futures
The technology must have the potential to be deployed at significant scale now or in the future and is a generation technology, a supporting technology or otherwise could significantly impact the generation sector. The broad categories that are currently considered relevant are:
•
Generation technologies
•
Storage technologies
•
Hydrogen technologies
•
Consumer scale technologies (e.g., rooftop solar PV, batteries).
Auxiliary technologies such as synchronous condensers, statcoms and grid-forming inverters are also relevant and important but their inclusion in energy system models is not common or standardised due to the limited representation of power quality issues in most electricity models. Where they have been included, results indicate they may not be financially significant enough to warrant inclusion. Also, inverters, which are relevant for synthetic inertia, are not distinct from some generation technologies which creates another challenge.
E.2 Transparent Australian data outputs are not available from other sources
Examples of technologies for which Australian data is already available from other sources includes:
•
Operating generation technologies (i.e., specific information on projects that have already been deployed)
•
Retrofit generation projects
•
New build transmission.
Most of these are provided through separate AEMO publications and processes.
GenCost 2025-26 | 103
Other organisations publish information for new build Australian technologies but not with an equivalent level of transparency and consultation. New build cost projections also require more complex methodologies than observing the characteristics of existing projects. There is a distinct lack of transparency around these projection methodologies. Hence, the focus of GenCost is on new build technologies.
E.3 Has the potential to be either globally or domestically significant
A technology is significant if it can find a competitive niche in a domestic or global electricity market, and therefore has the potential to reach a significant scale of development.
Technologies can fall into four possible categories. Any technology that is neither globally nor domestically significant will not be included anywhere. Any other combination should be included in the global modelling. However, we may only choose to include domestically significant technologies in the current cost update which is subcontracted to an engineering firm.
Apx Table E.1 Examples of considering global or domestic significance Globally significant Domestically significant Examples Yes Yes Solar PV, onshore and offshore wind
Yes
No
New large-scale hydro. No significant new sites expected to be developed in Australia
Conventional geothermal energy: Australia is relatively geothermally inactive No Yes None currently. A previous example was enhanced geothermal, but domestic interest in this technology declined
No
No
Emerging technologies that have yet to receive commercial interest (e.g., fusion) or have no commercial prospects due to changing circumstances (e.g., new brown coal)
E.4 Input data quality level is reasonable
Input data quality types generally fall into five categories in order of highest (A) to lowest (E) confidence in Australian costs:
A.
Domestically observable projects (this might be through public data or data held by engineering and construction firms)
B.
Extrapolations of domestic or global projects (e.g., observed 2-hour battery re-costed to a 4-hour battery, gas reciprocating engine extrapolated to a hydrogen reciprocating engine)
104 | CSIRO Australia’s National Science Agency
C.
Globally observable projects
D.
Broadly accepted costing software (e.g., ASPEN)
E.
“Paper” studies (e.g., industry and academic reports and articles).
While paper studies are least preferred and would normally be rejected, if a technology is included because of its potential to be globally or domestically significant in the future, and that technology only has paper studies available as the highest quality available, then paper studies are used. Confidential data as a primary information source is not used since, by definition, it cannot be validated by stakeholders. However, confidential sources could provide some guidance in interpreting public sources.
E.5 Mindful of model size limits in technology specificity
Owing to model size limits, we are mindful of not getting too specific about technologies but achieving good predictive power (called model parsimony). We often choose:
•
A single set of parameters to represent a broad class (e.g., selecting the most common size)
•
A leading design where there are multiple available (e.g., solar thermal tower has been selected over dish or linear Fresnel and single axis tracking solar PV over flat).
The approach to a technology’s specificity may be reviewed (e.g., two sizes of gas turbines have been added over time and offshore wind turbines have been split into fixed and floating). For a technology like storage, it has been necessary to include multiple durations for each storage as this property is too important to generalise. As it becomes clearer what the competitive duration niche is for each type of storage technology, it will be desirable to remove some durations. It might also be possible to generalise across storage technologies if their costs at some durations are similar.
GenCost 2025-26 | 105
Shortened forms Abbreviation Meaning AAS Australian Academy of Science
A-CAES
Adiabatic Compressed Air Energy Storage AE Alkaline electrolysis
AEMO
Australian Energy Market Operator ATSE Academy of Technological Sciences and Engineering
BAU
Business as usual BOP Balance of plant
CCS
Carbon capture and storage CCUS Carbon capture, utilisation and storage
CHP
Combined heat and power CIS Capacity Investment Scheme
CO2
Carbon dioxide CSIRO Commonwealth Scientific and Industrial Research Organisation
CST
Concentrated solar thermal EV Electric vehicle
FOAK
First-of-a-kind GALLM Global and Local Learning Model
GALLME
Global and Local Learning Model Electricity GALLMT Global and Local Learning Model Transport
GJ
Gigajoule GW Gigawatt
H2
Hydrogen hrs Hours
IAEA
International Atomic Energy Agency IEA International Energy Agency
ISP
Integrated System plan
106 | CSIRO Australia’s National Science Agency
Abbreviation Meaning kW Kilowatt
kWh
Kilowatt hour LAES Liquid Air Energy Storage
LCOE
Levelised Cost of Electricity LCOS Levelised cost of storage
LCV
Light commercial vehicle MCV Medium commercial vehicle
MLF
Marginal Loss Factor Li-ion Lithium-ion
LR
Learning Rate Mt Million tonnes
MW
Megawatt MWh Megawatt hour
NDC
Nationally Determined Contribution NEM National Electricity Market
NOAK
Nth-of-a-kind NSW New South Wales
NT
Northern Territory NZE Net zero emissions
O&M
Operations and Maintenance OECD Organisation for Economic Cooperation and Development
PEM
Proton-exchange membrane PHES Pumped hydro energy storage
PV
Photovoltaic REZ Renewable Energy Zone
SLCOE
System Levelised Cost of Electricity SMR Small modular reactor
STEPS
Stated Policies Scenario SWIS South-West Interconnected System
GenCost 2025-26 | 107
Abbreviation Meaning
TWh
Terawatt hour UAE United Arab Emirates
USC
Ultra-supercritical VPP Virtual Power Plant
VRE
Variable Renewable Energy WA Western Australia
WEM
Western Electricity Market WEO World Energy Outlook
108 | CSIRO Australia’s National Science Agency
References
Association for the Advancement of Cost Engineering (AACE). 1991, Conducting technical and economic evaluations – as applied for the process and utility industries, Recommended Practice No. 16R‐90, AACE International.
Australian Energy Market commission (AEMC) 2025, Residential electricity price trends 2025, AEMC
Australian Electricity Market Operator (AEMO) 2024, Integrated System Plan: A roadmap for the energy transition, AEMO.
Australian Electricity Market Operator (AEMO) 2025, Input and assumptions workbook, AEMO, https://aemo.com.au/consultations/current-and-closed-consultations/2025-iasr.
Australian Electricity Market Operator (AEMO) 2023b, 2023 Transmission expansion options report, AEMO.
Baik, E., Chawla, K. P., Jenkins, J. D., Kolster, C., Patankar, N. S., Olson, A., Benson, S. M., & Long, J. C. S. 2021, What is different about different net-zero carbon electricity systems? Energy and Climate Change, 2, 100046. https://doi.org/10.1016/j.egycc.2021.100046
CSIRO 2023, Renewable Energy Storage Roadmap, CSIRO.
Edmonds, J., Lucknow, P., Calvin, K., Wise, M., Dooley, J., Kyle, P., Kim, S.H., Patel, P., Clarke, L. 2013, Can radiative forcing be limited to 2.6Wm(-2) without negative emissions from bioenergy and CO2 capture and storage? Climatic Change, 118(1), 29-43 SI. doi:10.1007/s10584-012-0678.2
Energy Information Administration (EIA) 2017a, World installed liquids-fired generating capacity by region and country, US Department of Energy, Washington DC.
Energy Information Administration (EIA) 2017b, World installed natural-gas-fired generating capacity by region and country, US Department of Energy, Washington DC.
Energy Information Administration (EIA) 2023, Cost and Performance Characteristics of New Generating Technologies, Annual Energy Outlook 2023, US Department of Energy, Washington DC.
Fichtner 2026, the Australian concentrating solar thermal value proposition: Update summary, Prepared for ASTRI, https://www.astri.org.au/wp-content/uploads/2026/03/Fichtner-Final-Report-CST-Value-Proposition-2026-02-final.pdf.
Flyvbjerg, B. and Gardner, D. 2023, How Big Things Get Done: The Surprising Factors That Determine the Fate of Every Project, from Home Renovations to Space Exploration and Everything in Between, Crown Publishing.
Fraunhofer ISE, 2015. Current and Future Cost of Photovoltaics. Long-term Scenarios for Market Development, System Prices and LCOE of Utility-Scale PV Systems, s.l.: Study on behalf of Agora Energiewende
GenCost 2025-26 | 109
Gas Turbine World 2009, Gas Turbine World Handbook.
Gas Turbine World 2010, Gas Turbine World Handbook.
Gas Turbine World 2011, Gas Turbine World Handbook.
Gas Turbine World 2012, Gas Turbine World Handbook.
Gas Turbine World 2013, Gas Turbine World Handbook.
GHD 2026, 2025 energy technology costs and technical parameter review, AEMO.
Global Energy Monitor 2024a, Global Nuclear Power Tracker July 2024 release https://globalenergymonitor.org/projects/global-nuclear-power-tracker/
Global Energy Monitor 2024b, Global Coal Plant Tracker July 2024 release < https://globalenergymonitor.org/projects/global-coal-plant-tracker/>
Global Energy Monitor 2024c, Global Oil and Gas Plant Tracker August 2024 release < https://globalenergymonitor.org/projects/global-oil-gas-plant-tracker/>
Government of India. 2016, A new dawn in renewable energy. India: Ministry of new and renewable energy.
Graham, P. 2025, Least-cost modelling of electricity systems with a high share of weather dependent renewable generation, 26th International Congress on Modelling and Simulation MODSIM 2025, Adelaide.
Graham, P., Foster, J., Mediwaththe, C., Green, D. and Brinsmead, T. 2025, A simplified electricity system model for exploring least cost generation mixes, 26th International Congress on Modelling and Simulation MODSIM 2025, Adelaide.
Graham, P. 2018, Review of alternative methods for extending LCOE calculations to include balancing costs. CSIRO, Australia.
Grübler, A., Nakicenovic, N., & Victor, D. G. 1999, Dynamics of energy technologies and global change. Energy Policy, vol. 27(5), pp 247-280.
GWEC. (n.d.). Global Wind Report Series 2006 to 2016. Global Wind Energy Council.
Hayward, J. A., Foster, J. D., Graham, P. W., & Reedman, L. J. 2017. A Global and Local Learning Model of Transport (GALLM-T). In Syme, G., Hatton MacDonald, D., Fulton, B. and Piantadosi, J. (eds) MODSIM2017, 22nd International Congress on Modelling and Simulation. Modelling and Simulation Society of Australia and New Zealand, December 2017, pp. 818-824. ISBN: 978-0-9872143-7-9.
Hayward, J.A. and Graham, P.W. 2017, Electricity generation technology cost projections: 2017-2050, CSIRO, Australia.
Hayward, J. and Graham, P. 2013, A global and local endogenous experience curve model for projecting future uptake and cost of electricity generation technologies, Energy Economics, 40, 537-548.
Hayward, J. and Graham, P. 2011, Developments in technology cost drivers – dynamics of technological change and market forces, CSIRO. http://hdl.handle.net/102.100.100/104795?index=1
110 | CSIRO Australia’s National Science Agency
Idel, R. 2022, Levelized full system costs of electricity, Energy, vol. 259, pp. 1-11.
International Energy Agency (IEA) 2008, World Energy Outlook. Paris, France: IEA.
International Energy Agency (IEA) 2015, 2015 Projected costs of Generating Electricity, Paris, France: OECD.
International Energy Agency (IEA) 2016a, CSP Projects Around the World. Retrieved from SolarPACES: http://www.solarpaces.org/csp-technology/csp-projects-around-the-world
International Energy Agency (IEA) 2016b, World Energy Outlook, Paris, France: OECD.
International Energy Agency (IEA) 2019, Nuclear power in a clean energy system, IEA OECD.
International Energy Agency (IEA) 2021, World Energy Outlook, Paris: OECD.
International Energy Agency (IEA) 2022, World Energy Outlook, Paris: OECD.
International Energy Agency (IEA) 2023, World Energy Outlook, Paris: OECD.
International Energy Agency (IEA) 2024a, World Energy Outlook, Paris: OECD.
International Energy Agency (IEA) 2024b, Global Hydrogen Review 2024, Paris: OECD.
International Energy Agency (IEA) 2025, World Energy Outlook, Paris: OECD.
International Energy Agency (IEA) 2026, World Energy Investment 2026, Paris: OECD.
International Renewable Energy Agency (IRENA) 2023, Renewable Power Generation Costs in 2022, International Renewable Energy Agency, Abu Dhabi.
International Renewable Energy Agency (IRENA) 2024, Renewable Energy Statistics 2024, International Renewable Energy Agency, Abu Dhabi.
ITP Thermal 2024, CSP modelling for Gencost LCOE: Discussion paper, Australian Solar Thermal Research Institute.
Lazard 2023, Lazard’s Levelized Cost of Energy Analysis—Version 16.0, Lazard.
McDonald, A., and Schrattenholzer, L. 2001, Learning rates for energy technologies. Energy Policy, vol. 29, pp 255-261.
Mott MacDonald 2023, AEMO transmission cost database, building blocks costs and risks factors update: Final report, AEMO.
Neij, L. 2008, Cost development of future technologies for power generation-A study based on experience curves and complementary bottom-up assessments. Energy Policy, 36(6), 2200-2211.
Oxford Economics Australia 2025, 2025 IASR Planning and installation cost escalation factors: Report for AEMO, AEMO, Melbourne.
Rubin, E. S., Yeh, S., Antes, M., Berkenpas, M., & Davison, J. (2007). Use of experience curves to estimate the future cost of power plants with CO2 capture. International Journal of Greenhouse Gas Control, vol. 1(2), pp 188-197.
Samadi, S. 2018, The experience curve theory and its application in the field of electricity generation technologies, Renewable & Sustainable Energy Reviews, Vol. 82, pp 2346-2364
GenCost 2025-26 | 111
Schmidt, O., Hawkes, A., Gambhir, A., & Staffell, I. 2017, The future cost of electrical energy storage based on experience rates. Nature Energy, vol. 2, 17110.
Schoots, K., Kramer, G.J. and Zwaan, B. 2010, Technology Learning for Fuel Cells: An Assessment of Past and Potential Cost Reductions, Energy Policy, vol. 38, pp. 2887-2897
Stehly, T., and Duffy, P. 2021, 2021 Cost of Wind Energy Review, NREL, https://www.nrel.gov/docs/fy23osti/84774.pdf
UN 2015a, Energy Statistics Yearbook 2012. New York, USA: United Nations.
UN 2015b, Energy Statistics Yearbook 2013. New York, USA: United Nations.
US Department of Energy (DOE) 2024, Pathways to commercial nuclear liftoff: Advanced nuclear, DOE.
Voormolen, J.A., Junginger, H.M. and van Sark, W.G.J.H.M., 2016. Unravelling historical cost developments of offshore wind energy in Europe, Energy Policy, Elsevier, vol. 88(C), pp 435-444.
Wilson, C. 2012. Up-scaling, formative phases, and learning in the historical diffusion of energy technologies. Energy Policy, 50(0), 81-94, http://dx.doi.org/10.1016/j.enpol.2012.04.077
Wiser, R., Rand, J., Seel, J., Beiter, P., Baker, E., Lantz, E. and Gilman, P., 2021. Expert elicitation survey predicts 37% to 49% declines in wind energy costs by 2050, Nature Energy, vol. 6, pp 555-565.
Wright, T. P. 1936, Factors Affecting the Cost of Airplanes. Journal of the Aeronautical Sciences, vol. 3, pp 122-128.
Zwaan, B., Rivera Tinoco, R., Lensink, S. and Van den Oosterkamp, P. 2012, Cost reductions for offshore wind power: Exploring the balance between scaling, learning and R&D, Renewable Energy, vol. 41, pp 389-393
112 | CSIRO Australia’s National Science Agency
The GenCost project is a partnership of CSIRO and AEMO.
As Australia’s national science agency and innovation catalyst, CSIRO is solving the greatest challenges through innovative science and technology.
CSIRO. Unlocking a better future for everyone.
Contact us
1300 363 400
+61 3 9545 2176
www.csiro.au/en/contact
For further information
Energy
Paul Graham +61 2 4960 6061
paul.graham@csiro.au
csiro.au/energy