As I write, the National Dialogue is in progress. While there is little doubt that South Africa’s future needs to be reset through dialogue, it is essential that it does not get detached from the hard realities and tough challenges we face as a nation. Trust does not get rebuilt with grand statements about a shared future as inequalities get worse; it gets built by making decisions that result in actual improvements in the quality of daily life. This is particularly true of our energy challenges. Energy affects the daily life of every South African.
No one will disagree that load shedding has not only crippled our economy since the power crisis began in 2008; it also engendered an atmosphere of disillusionment, despair and even helplessness. After all, there are kids finishing school now who have never had a year without load shedding. For them, dysfunction has been normalised. And yet, we expect them to be hopeful about the future.
The tide, of course, turned in mid-2022 when President Cyril Ramaphosa announced the Energy Action Plan (EAP), the appointment of Kgosientsho Ramokgopa as (initially) Minister of Electricity in the Presidency (confirmed after the May elections as Minister of Energy and Electricity) and the establishment of the National Electricity Crisis Committee. For the first time since 2019 we had a minister who was prepared to admit that the causes of load shedding had much to do with the poor performance of our coal-fired power stations and not the treasonous intentions of the Eskom leadership. This created the context for solutions, including investing in improved performance and aggressively supporting the expansion of renewables.
Although challenges remain, there is a sense that the energy crisis is over. Attention, therefore, has turned to the medium- to long-term future. Big questions remain unanswered: What will the energy mix be? What is the lowest-cost option? How do we decarbonise the South Africa economy? How much is needed to pay for the infrastructures that must replace our ancient fleet of coal-fired power stations? Where will the money come from?
Like all long-term future planning in the computer age, this means building complex computerised models that make it possible to construct future scenarios and then assessing – based on very detailed empirical research – which scenarios are likely if nothing changes; as well as those more preferred scenarios that will require decision-makers to make decisions that will determine what happens over the next few decades through to 2050 and beyond. In short, our quality of everyday life in the future will depend on how we interpret the models. The stakes, therefore, could not be higher.
Unlike most other Global South countries, South Africa is well endowed with several modelling capabilities located within Eskom, universities and the private sector. This is very good news, because it means debate and discussion about future options can be informed by an array of modelling outcomes rather than just one (or worse, reliance on an externally developed model as often happens in the Global South).
In 1976, George E.P. Box, a British statistician, coined the well-known phrase: “All models are wrong, but some are useful.” What this refers to is that, contrary to popular opinion, models are not about the truth. Nor do they predict the future. They are only useful for enriching democratic deliberation about future options.
The decisions that get taken should not be determined by the model, but rather by the judgements shaped through democratic deliberation and debate. This is why South Africa is so fortunate; our deliberations benefit from the outcomes of a range of modelling processes undertaken by competent researchers aligned with diverse interests and motivated, in certain respects, by different values. They all take evidence very seriously when they build their models, but it is the differences in how they interpret the evidence based on their assumptions about future trends that generate often quite different outcomes.
Over the next two months, two reports will be published for public debate that contain the results of two major modelling initiatives. The first is the 2025 iteration of the Integrated Resource Plan (IRP), revised following expert feedback on the 2024 draft. IRP 2025 is expected to be published by the end of August or early September after approval by Cabinet for discussion. It has been extensively discussed with a vast range of stakeholders, including Nedlac.
The second is a report that will be launched in early October titled South Africa’s Energy Sector Investment Requirements to Achieve Energy Security and Net-Zero by 2050* (referred to below as the SAESIR report) that has been carefully developed over the past two and a half years by a joint team from the National Planning Commission, SA-TIED, Presidential Climate Commission (PCC) and the Development Bank of Southern Africa (DBSA), with funding provided by the DBSA and the PCC*. Both reports are expected to take seriously the need to decarbonise the South African economy in line with Chapter 5 of the National Development Plan (NDP) and South Africa’s Nationally Determined Contribution (NDC).
In a 2021 paper by University of Cape Town (UCT) researchers commissioned by SA-TIED, no less than 13 previous studies are reviewed that model future low-carbon development pathways for South Africa. This UCT-based modelling group has compiled the most credible estimation of future energy demand through to 2050, which most modelling initiatives (including the IRP) have incorporated. Their conclusion in their SA-TIED paper is that a low-carbon energy transition is also the lowest cost. Since then, most models have attempted to reconcile what is referred to as the “lowest-cost” solution, which also ensures decarbonisation (sometimes referred to as “net zero”). Some focus on total energy requirements of the whole economy (like UCT’s SA-TIED paper), while others just focus on the electricity sector.
The Presidency’s Just Energy Transition – Investment Plan (JET-IP) emphasises the electricity sector, but brings in wider energy sectors (such as green hydrogen and electric vehicles). The JET-IP estimated that the total investment required for the period 2023-2027 would be R1.4-trillion (annually nearly R300-billion) to achieve the goals of the NDC, NDP and the “net zero” target.
Taking into account the need to decarbonise the entire economy at lowest cost, the PCC estimated that the total investment required through to 2050 would be R8.6-trillion (just over R300-billion per annum). The National Business Initiative (NBI) estimated in 2022 that the total investment requirement to achieve the goals of the NDC and net zero by 2050 was R5.9-trillion for the period 2020-2050 (nearly R200-billion annually).
The Blended Finance Task Team (in collaboration with the Centre for Sustainability Transitions at Stellenbosch University) estimated in 2022 that the total investment requirement to decarbonise only the electricity sector at lowest cost would be R3.8-trillion to achieve the goals of the NDC and the “climate action” Sustainable Development Goal (SDG) (R136-billion per annum). This report included cost estimates for extending the transmission grid.
Eskom’s 2022 estimate was that R1.2-trillion would be required for the period 2023-2035 to fund its Just Energy Transition, including investments in generation, transmission and distribution. This did not include the current estimate that R400-billion is required to extend the transmission grid by 14,400km and 210 substations over the next decade.
The World Bank’s 2022 Climate and Development Report for South Africa estimated that R2.4-trillion would be required over the period 2022-2030 to fund the triple transition — namely the low‑carbon transition, climate adaptation/resilience, and ensuring a just transition.
In other words, estimates of annual investment requirements differ according to scope of investment, i.e. electricity sector only, or economy-wide energy consumption. That said, estimates range from R150-billion to R300-billion per annum.
The IRP and SAESIR reports will more than likely focus on the electricity sector rather than economy-wide decarbonisation. Both are likely to recommend lowest-cost options that result in decarbonisation (i.e. coal closure over time). And both are likely to endorse the current commitment to building 4-5GW per annum of renewable energy (including extensive battery backup) as well as the ambitious R400-billion Transmission Development Plan to rehabilitate and extend the transmission grid. Indeed, both will be influenced by the fact that an estimated 30GW of renewable energy generation capacity is already in the pipeline, which means we are ahead of the 2030 targets for renewables that have been proposed by most energy models to date.
The differences, however, may well emerge when it comes to nuclear power and gas. The choices in this regard will be influenced by whether or not the researchers assume that “baseload energy” is still an absolute necessity or not.
“Baseload energy” refers to the assumption that modern electricity systems need a non-variable supply of energy which is what coal-fired or nuclear power stations provide. Coal-fired power stations need to run at a constant output, but can only be adjusted up or down relatively slowly. They therefore need to be complemented by generators that can instantly dispatch energy on demand, normally gas-powered “peakers”.
From a “baseload” perspective, renewables plus battery backup (for gaps measured in hours) and gas backup (for gaps measured in days) will not ensure energy security. It follows, that if coal closure is assumed, “baseload” can only come from nuclear power and/or gas (used for baseload, not just backup).
However, the assumption that ‘baseload” energy is required to ensure energy security in a renewables-based economy has in recent years been seriously questioned in light of the capabilities made possible by AI and so-called smart grids. In essence, from this point of view, energy planning has shifted from a state-controlled supply-driven paradigm (driven by the long lead times and high costs of building coal-fired power stations, which means future demand must be assumed) to a more market-based demand-driven paradigm (made possible by very short lead times to build renewables in response to real shifts in demand plus the advantages of “smart grid” and AI technologies).
There is much interest in how the IRP 2025 will treat the question of nuclear power. Some have speculated that it might echo recent statements by the Minister of Energy and Electricity, Globally, the role of nuclear in energy transitions is debated, particularly given questions of costs and construction timelines. South Africa’s planning will need to weigh these considerations carefully in light of local investment realities and recent expert evidence. Given that the CO2 footprint of nuclear power is similar to solar power (and not zero, as articulated by the pro-nuclear lobby), it follows that nuclear will be presented as consistent with NDC targets and Chapter 5 of the NDP.
What will really matter is the underlying assumptions made about the levelised cost of energy (LCOE)* for nuclear power, the length of the assumed life cycle and how long it will take to build a nuclear power plant. Obviously, the longer the life cycle, the lower the LCOE of nuclear. Many, moreover, will question whether a nuclear power plant will take less than a decade to build, which takes us into the late 2030s before realising the value of nuclear power. What happens until then?
Lazzard’s Levelized Cost of Energy Report is the most authoritative source for costs of energy technologies. The 2025 report shows that the cost of (unsubsidised) solar PV has on average dropped slightly from a 2023 range of $29-$92/MWh relative to a 2024 range of $33-$44/MWh; onshore wind has shifted from a 2023 range of $27-$73/MWh to a 2024 range of $37-$66/MWh; coal has risen from a 2023 average range of $69-$168/MWh to a 2024 average range of $109-$157/MWh; and gas peaking costs have dropped from a 2023 range of $110-$109/MWh to a 2024 range of $48-$109/MWh. Nuclear power costs have risen significantly from a 2023 range of $142-$222/MWh to a 2024 range of $141-$251/MWh. In other words, nuclear is a lot more expensive than solar and wind power (including backup).
It will be interesting to see how the IRP interprets these figures. More than likely LCOE will be criticised by the IRP for excluding the related costs of renewables, such as grid transmission extension and backup. Hopefully, this will be balanced against the related costs of nuclear, in particular the well-known construction delays and insurance costs associated with nuclear. Our failure to build Medupi and Kusile power stations on budget and on time should serve as a big flashing red light when it comes to considering future mega projects.
A non-baseload perspective that leaves out nuclear and builds a lot of renewables will have to worry about backup to what is called “variable renewable energy” (VRE). There are two challenges here: gaps in solar and wind generation measured in hours, and gaps measured in days (say three to four days). For the former, you need a lot of battery backup for instant dispatchability. For the latter, you need substantial gas infrastructure (i.e. capacity measured in GW), even though it will lie idle for most of the time (i.e. not a lot of gas will be used to generate electricity measured in GWh). The Blended Finance Task Team report estimated that 30GW of gas backup will be required for a VRE-based grid.
Different modelling exercises, even if they use the same software, can generate different outcomes depending on assumptions made and interpretations of the cost estimates. For example, for a model to select a lot of nuclear without suppressing renewables build, it will be necessary to significantly limit gas production. Similarly, estimates of costs of nuclear could result in high or low costs per MWh depending on assumptions about the lifespan of nuclear power plants. The longer the lifespan, the lower the average cost per MWh.
By contrast, leaving out nuclear for cost reasons and selecting large quantities of gas to back up VRE has its own risks. There is little evidence that there is significant investment appetite for gas infrastructure, in particular if used for backup and not baseload. This could change if gas infrastructures were treated like batteries, i.e. monthly payments for availability irrespective of output. That would create the certainty needed to secure debt and equity financing.
Finally, given that government has repeatedly emphasised that the fiscal capacity of the state is too constrained to finance the energy transition, it follows that substantial quantities of private sector funding will be required. If this is true, then much will depend on cost of capital (i.e. interest payable on the funds accessed from private funders).
In its 2024 report, Lazzard assessed the differential impact of cost of capital on different technologies. It used two scenarios: the low-cost scenario assumed a WACC* of 5.4%, cost of equity at 8% and cost of debt at 6%; the high-cost scenario assumed a WACC of 10%, cost of equity of 16% and a cost of debt at 10%. When applied to nuclear and solar PV (utility scale), the cost of nuclear power in the high-cost scenario rises by 52% compared with the low-cost scenario (i.e. from $148/MWh to $226/MWh). By contrast, the cost of utility scale solar PV in the high-cost scenario only rises by 23% compared with the low-cost scenario (i.e. from $69/MWh to $85/MWh).
When the IRP 2025 and SAESIR reports come out we need to scrutinise the assumptions made about costs. To be realistic, they both need to (a) consider capital and operating costs over the life cycle; (b) the inclusion of the costs of both generation and transmission/distribution; and (c) also take into account the capacity of private funders to fund energy infrastructure through to 2050, including the cost of this capital.
Finally, the international landscape needs to be considered strategically. By 2022, fossil fuels accounted for 61% of total energy consumption, compared with renewables, which was 29%, and nuclear was 9% (436 reactors in 37 countries). Total energy consumption was 29PWh. The International Renewable Energy Association (Irena) has estimated that by 2030 total consumption will have risen to 40PWh, with 24% sourced from fossil fuels, 68% from renewables and 8% from nuclear. Their 2050 scenario is total consumption rising to 90PWh, with 4.4% sourced from fossil fuels, 4.2% from nuclear and 91.4% sourced from renewables (including, of course, massive backup systems from batteries, gas, hydro, etc).
South Africa needs to position itself within this fast changing international context, but also have a strategic eye on the African market. Africa’s total generation capacity in TWh is less than France and Germany combined. Compared with France and Germany’s total population of 140 million, Africa’s population is well over a billion and many economies are growing faster than most other economies in the world.
Eleven of the 20 countries on the IMF’s list of the 20 fastest-growing economies in 2024 were African economies. It, therefore, follows that if Africa energises using fossil fuels, none of the climate targets approved in Paris in 2015 will be achieved. The world, therefore, has an interest in funding low-carbon development in Africa. Will we become Africa’s largest supplier of the kit needed to make this happen, or will we leave this to China? Our industrial future depends on the answer to this question.
We need to build an energy transition strategy that assumes that (in a post-Trump era) the world will respond to the ever-worsening climate crisis by accelerating the transition to net zero by 2050. We also need to assume (based on all the published models) that the least cost option for South Africa is also about gradual coal closure over time (with around 15GW decommissioned by 2030) and massive growth in VRE backed up by batteries and gas. The returns on (both public and private) investment internationally and in South Africa are more attractive than the alternatives, which is why it is likely to happen. The construction and operational risks are, of course, also lower.
This is the framework that should be used to assess the two upcoming modelling-based reports, namely the IRP 2025 from the Department of Energy and Electricity and the SAESIR Report from the collaboration between the NPC/PCC/DBSA/SA-TIED. If the National Dialogue does not address this challenge in a bold, clear-eyed manner, the quality of daily life could be compromised for many years to come. The models that underpin these two reports will help enrich what should be a national dialogue about our energy future. DM
Disclaimer: Mark Swilling writes in his academic capacity and, therefore, this article reflects his personal view and neither the views of any of the parties involved in the two modelling initiatives he cites, nor the National Planning Commission.
*Footnotes
*The SAESIR report will be publicly launched in early October.
*The professional team that did the work was led by PriceWaterhouseCoopers, and included Osmotic Engineering Group who did the energy modelling work. A Steering Committee comprised of representatives from NPC, PCC, DBSA and SA-TIED and Chaired by Prof Sampson Mamphweli from SANEDI provided strategic guidance.
*LCOE is total lifetime costs (capex and opex) divided by total energy produced over the life cycle.
*Weighted Average Cost of Capital.