Fermi, on firming the grid: 'Are you all crazy?'

The information here was gathered around the end of January 2026 by direct interrogation of GenCost using ChatGPT. Two costs stood out. Capital cost of a generating system depends on size and electrical load, which need specification. Interestingly, ChatGPT says it cannot specify "a single published figure" for Sydney's maximum load. But it does quote a range as typical, 18 – 23 GW (gigawatts). And it gives a single figure from GenCost of AUD 1.3 million per MW for generating plant installation/construction costs. With these numbers the range of total costs (rounded) becomes AUD 23 – 30 billion. GenCost explicitly excludes some cost components, like "land, grid connections and transmission upgrades". But what it does include is a good start.

Firming is the other major item. It adds to cost but greatly expands utility of a system. According to ChatGPT, lithium-ion batteries "are the mainstream choice today for grid-scale projects". I don't dispute that claim but it does need qualification. Lithium batteries have revolutionised motor cars and portable electrical equipment like tools. Their key advantage is high electrical energy density. But they are expensive and the more mundane performance needs of stationary energy storage should in principle suit cheaper, lower energy density batteries. Clearly the power generation industry values some other benefits lithium batteries offer, like reliability. So, lithium it is.

Two parameters are required to specify firming performance, delivery rate (power capability) and cycle life (how long the charged battery needs to keep supporting the grid). For the Sydney example ChatGPT specifies a power output of 18 – 23 GW for up to 6 hours. This translates to a required storage capacity of up to 108 – 138 GWh, at a cost of $AU 62 – 80 billion. Potential additional costs, like grid connection, inverters, land, project financing, permitting and operating costs, are mentioned but not estimated.

In summary, a utility-scale solar PV-based clean generation system for Greater Sydney, coupled with a lithium battery-based firming system, is costed by GenCost at between $85 billion and $110 billion. The latter figure comprises $30 billion for generation and $80 billion for firming.

These numbers demonstrate the overwhelming effect of firming on cost; at the upper end of cost estimates, firming by batteries represents nearly three-quarters of the total cost $110 billion for a reliable large solar/wind-based power supply suited to a major city.

Who knew? Fermi probably would have. A physicist will appreciate that traditional fuel-driven power sources come with their own storage since fuels by definition are stores of energy. That energy is released as heat on combustion. Hot steam follows, which in the power plant is the medium by which heat energy drives spinning generators.

Solar and wind energy are not fuels. The extra costs of firming they require could turn out to be a dealbreaker, despite the current popularity of renewables. Fermi, a pioneer of clean nuclear energy that needs no firming at all, obviously would have understood this. Which is why I suggest he would have greeted the above calculations and cost figures with "Are you all crazy?"

There's a new case for nuclear energy – it uses a fuel! Time to look at it again.