Why nobody actually knows if nuclear is cheaper

GenCost prices each technology on its own (LCOE: the average lifetime cost of electricity from one kind of plant), and from 2025-26 adds a system-wide version (SLCOE) that counts the cost of keeping variable power reliable. Frontier refuses the per-technology frame and prices whole pathways instead. Each side's headline answers only its own question, so the two numbers were never comparable to begin with.

The line that moves the most money. Frontier's like-for-like run, nuclear inside the same demand future AEMO uses, comes to 25 per cent cheaper. The campaign's 44 per cent came from pairing a different demand future with the technology change, a pairing GenCost's authors ruled uninterpretable. Nearly half the political claim lived in that choice alone. The climate question hides here too: Frontier says its modelling meets net zero by 2050 with lower average emissions sooner than AEMO's plan; GenCost's stated reason for ruling the comparison out is that the savings bundle "delayed emissions reduction, lower economic growth and increased nuclear generation", three factors it says cannot be separated.

The dispute underneath is which past best predicts an Australian reactor: Australian coal, or the world's nuclear fleet. Each model settles it by assumption, not evidence. One provenance note: the 90 per cent figure never appears in Frontier's nuclear report itself. It runs through the modelling the report mirrors and through Frontier's public defence, which is why it wears the supplied star here. GenCost prints its range as 53 to 89; the 54 in Grattan's quote below is his rounding, not CSIRO's.

The longer the recovery window, the cheaper each year of nuclear looks. Neither report argues for its window; each asserts one. GenCost's reply on long life: wind and solar sites can also run 50 to 60 years by rebuilding at falling cost, capturing a similar benefit.

GenCost holds that a first reactor is at least 15 years away, a clock that starts at repeal of a ban still in force. Frontier keeps most of the coal fleet running to bridge the wait, and answers the feasibility worry by noting AEMO's closure schedule runs well ahead of the owners' own announced dates. The fuller the bridge, the cheaper the pathway looks.

Both models use similar build costs for an established plant. The fight is the first one. Frontier confines its costs to equilibrium, meaning a plant built once the industry already exists. GenCost says global experience puts the first plant at more than double, and Australia has no industry to learn from. Whether the learning arrives before or after the bill does a large share of the work in both conclusions.

Opportunity cost is what the money could earn elsewhere: a model prices it as a discount rate, and the higher the rate, the less a saving in 2050 is worth today. GenCost discloses its rate and the reason for it. Frontier's report leaves its rate behind a broken footnote, so the 25 per cent margin cannot be recomputed by a reader. One side can be checked; the other has to be trusted.