← Energy Realism

The Real Cost

A$173bnand ~20 million tonnes of material have gone into the renewable build. Here is what that could have bought instead — and the bills the headline figure doesn't capture. Every number is sourced and the honest caveats are shown, because a number that can be knocked down is worthless.

1 · The nuclear counterfactual

The same A$173bn in nuclear
1117 GW
delivering 80150 TWh/yr at 90% capacity factor — around Australia's entire current renewable output, from plants that run 60 years.
What the renewable fleet delivers today
97 TWh/yr
all renewables in 2024 (wind + solar = 84TWh), from panels & turbines replaced every ~25–30 years. Same money, a fraction of the working life.
A$10bn–A$16bn
Cost / GW (AU FOAK)
90%
Capacity factor
60 yr
Plant life
~2040
First power

The honest catch — lead time, not cost, is the real problem. At a realistic Australian first-of-a-kind cost (~A$10–16bn/GW overnight), A$173bn buys ~11–17 GW of nuclear — delivering ~80–150 TWh/yr at 90% capacity factor (around today's entire renewable output), from plants that run 60 years. But it would arrive from ~2040 at the earliest. Overnight cost only (excludes interest-during-construction, grid, decommissioning). You cannot honestly quote Korea/China fleet prices for a zero-base build. The strongest counterfactual isn't the dollar figure (CSIRO assumes a 120% first-of-a-kind premium and a ~59% load-following capacity factor, which would cut delivered TWh ~35%) — it's that the first reactor couldn't deliver power before ~2040, 15+ years after the ban is lifted. Sources: CSIRO GenCost 2024-25, IAEA, World Nuclear Association, Lazard.

2 · The bill the nameplate doesn't pay

53 GW
wind + solar nameplate already installed
~1015%
of that counts as firm capacity at the peak
74 GW
dispatchable fleet still needed by 2050 (~4× today)
To meet a ~35 GW peak and ride out "dark & still" weeks, AEMO's plan builds 49 GW / 646 GWh of storage + 15 GW of flexible gas + 7 GW hydro— a firm fleet that exists almost regardless of how much wind & solar nameplate is built.

Stated honestly: Variable renewables don't inherently supply firm capacity: ~53 GW of installed wind+solar counts for only ~5–21% of nameplate at the coincident peak (~35 GW) — solar's credit falls toward ~5% at saturation, wind ~11–21% (NREL ELCC). So AEMO's plan (2024 ISP, Step Change scenario) builds a SEPARATE ~74 GW dispatchable fleet by 2050 — ~4× today — that must be built and paid for on top. Honest: that gap can be filled by storage, hydro, gas OR nuclear — it is not 'the same fossil fuels' — but it is an unavoidable ADDITIONAL system cost the headline figures don't capture. Sources: AEMO 2024 ISP (Step Change), NEM reliability standard, NREL ELCC, CSIRO GenCost integration costs.

3 · Lifecycle carbon — both scenarios

Coal (status quo)
820 g
Gas (status quo)
490 g
Renewable system — Australia-realistic
65 g
Renewable system — best-case headline
40 g
Nuclear system
20 g
grams CO₂ per kWh, lifecycle · the Australia-realistic figure can reach ~145 g if firming is gas-heavy
Why the headline understates it
The rosy headline
Already cradle-to-grave — mining, refining, transport, manufacture and waste are ALL inside this number (that is what 'lifecycle' means). But it assumes panels made on a cleaner grid, a single lifecycle, and no firming.
~40 g/kWh
+ Made on China's coal grid
Chinese polysilicon & wafer manufacture carries ~1.3–1.7× the emissions of a cleaner grid (Fraunhofer ISE; Argonne) — and ~95% of panels run through it.
~50–60
+ Firmed with gas
Back renewables with gas and the backup's operating emissions drag the whole system up. Australia's plan keeps ~15 GW of gas for deep-firm insurance.
up to ~145
+ Rebuilt 2–3× per reactor life
Over a reactor's 60-year life the renewable fleet is remanufactured 2–3 times — so its factory emissions keep recurring, while a reactor's are largely spent once, up front.
total, not intensity

Both grids beat coal/gas (~600–820 g/kWh) — conceded openly. The honest catch: the rosy ~40 g/kWh headline already includes the whole supply chain (you can't add mining/refining/transport on top — that double-counts), but it's a best case — clean-grid manufacture, one lifecycle, no firming. Australia's REAL case — China-coal-made panels plus gas firming — lands realistically around ~65 g/kWh (and up to ~145 if gas-heavy): several times the headline, well above nuclear's ~15–25, though still below gas. And because the fleet is rebuilt 2–3× over a reactor's life, its TOTAL emissions keep accruing. The gap is driven by firming choice + manufacturing location — not any inherent renewable virtue. Mt-CO₂ totals are illustrative; the ratio and drivers are the robust takeaway. Sources: IPCC AR5 Annex III, UNECE 2021/22 lifecycle, CSIRO GenCost, Fraunhofer ISE, Argonne.

4 · Waste — two axes, both shown

By mass
~2.5 Mt
solar-panel waste by 2050 (glass + aluminium, <15% recycled), plus ~30kt of landfilled wind blades.
vs nuclear spent fuel: ~60kt over a 60-year fleet — 2–3 orders of magnitude less by mass.
By hazard
The mirror image
Nuclear waste is intensely radioactive and long-lived — but tiny, fully inventoried, and engineered-for (pools → dry casks → geological disposal). Renewable waste is largely inert and non-radioactive — but vast, mostly landfilled, with localised battery-fire and leachate risks.

Two axes. By MASS: renewable end-of-life waste outweighs nuclear spent fuel by ~2–3 orders of magnitude — millions of tonnes of PV glass/aluminium by 2050, <15% recycled, mostly landfilled. By HAZARD: nuclear spent fuel is intensely radioactive and long-lived but tiny (~60kt over a 60-yr fleet), fully inventoried and engineered-for. Both end-states are unfinished — no deep geological repository operates yet; most panels and blades are landfilled. Sources: IRENA End-of-Life PV, Salim et al. 2023 (AU PV waste), DCCEEW, World Nuclear Association.

Both a renewable grid and a nuclear grid beat coal and gas decisively on carbon. The case here is narrower and sourced: the renewable path costs more in dollars, material tonnage, China-dependence, firming and end-of-life waste-by-mass than its advocates count — and a firm, long-lived, low-material option was ruled out by law. Read the full methodology & every caveat in the repo's REAL_COST_RESEARCH.md.