I spent part of last weekend at a chess club afternoon in Brunswick, the kind of slow, low-stakes tournament where nobody’s rated above 1800 and everyone argues about the clock more than the position. Somewhere around move thirty in a drawn-out endgame, it struck me that the gas-versus-batteries argument in the NEM has the same flaw as a lot of amateur chess: people keep comparing pieces as if they do the same job. A rook and a bishop both move fast and look useful on an open board. A 250 MW gas peaker and a 250 MW battery both look useful on a capacity chart. Neither comparison tells you much until you ask what the position actually needs.
That’s the frame I want to use today, because the “gas peakers versus big batteries” debate keeps getting flattened into a cost-per-megawatt argument when the honest version is a cost-per-job argument. The jobs are different. Sometimes gas wins. Increasingly, batteries win. And there’s a real, specific reason AEMO’s own planning documents keep both in the mix rather than picking a winner outright.
What firming actually means in practice #
We’ve covered the definition before in What ‘firming’ means and why gas keeps coming up, but it’s worth restating plainly: firming is the capacity that shows up when wind and solar don’t. It’s not baseload, it’s not peak demand in the traditional sense. It’s the gap-filler, and the size of the gap varies from twenty minutes to several days.
That variance is the whole ballgame. A battery with two hours of storage is brilliant at smoothing a sharp evening ramp, the kind that happens every day as rooftop solar falls away and air conditioners stay on. It is close to useless for a three-day wind lull across south-eastern Australia, the sort AEMO’s seasonal outlook flags most winters. Gas has the opposite profile: slower to respond in sub-five-minute terms, but it can run for as long as the pipeline keeps delivering fuel.
Power versus energy, said properly this time #
Let’s be careful with that number, because this is where most comparisons go wrong. A gas peaker rated at 250 MW and a battery rated at 250 MW are not interchangeable assets, because megawatts describe power – how fast energy moves – not how much energy exists to move. The battery’s second number, its storage duration, is the one that actually determines what it can do.
Take a 250 MW battery with two hours of storage. That’s 500 MWh. Run it flat out and it’s spent by the time the evening peak is half over. A 250 MW open-cycle gas turbine with a full fuel contract can, in principle, run for days – limited by gas supply and emissions constraints, not by a state-of-charge gauge. So when you see a headline comparing “250 MW of batteries” to “250 MW of gas” as if they’re equivalent firming capacity, that’s the point to stop and ask for the duration figure. AEMO’s Electricity Statement of Opportunities treats this distinction as central, reporting both installed capacity and expected energy contribution separately for exactly this reason.
A worked example
Here’s a simple table that I use when I’m trying to explain this to someone outside the industry: a council planner, say, or a bank credit officer sizing a loan against a project.
- Two-hour battery, 100 MW / 200 MWh: covers a sharp evening ramp, roughly 6pm to 8pm, cheaply and instantly. Can’t help with a multi-day lull.
- Four-hour battery, 100 MW / 400 MWh: covers most of the evening peak and into the night. Becoming the new standard for projects clearing the Capacity Investment Scheme, as we discussed in the CIS tender clock piece.
- Open-cycle gas turbine, 100 MW, no storage limit: can run for hours or days depending on gas supply, but takes longer to start from cold and burns fuel the whole time it’s running.
None of those three is “better” in the abstract. They answer different questions about how long the gap lasts.
The economics nobody wants to say plainly #
Gas peakers are expensive to run and cheap-ish to build relative to their nameplate capacity, assuming the site already has a gas connection. Batteries are the reverse: heavier upfront capital cost per megawatt, but once built, the marginal cost of dispatching stored energy is close to the price of the charging electricity, which, on a day with abundant rooftop and utility solar, can be close to zero or even negative.
That’s why batteries have been eating the short-duration end of the firming market so fast. The economics of the five-minute dispatch window, which we’ve gone through in detail in how the NEM actually dispatches power every five minutes, reward an asset that can start and stop instantly and chase price spikes without burning fuel. Projects like the Waratah Super Battery in NSW exist largely because of that speed advantage, not primarily as an energy source but as a system stabiliser, a role we unpacked in Waratah Super Battery: the shock absorber finally earning its keep.
Gas keeps its place at the longer end. Kurri Kurri in the Hunter Valley is the clearest recent case study: a project that took years longer than planned and arrived into a market where batteries had already claimed a lot of the short-duration revenue it was originally designed to chase, a point I went into in Kurri Kurri: the peaker that took the long way round. It still has a job. It’s just a narrower one than originally sold.
Where AEMO’s own numbers put the split #
AEMO’s Integrated System Plan, which we’ve broken down line by line in what the AEMO Integrated System Plan actually says, models a NEM that needs both dispatchable gas and large-scale storage well into the 2030s and beyond, with gas increasingly running at low capacity factors – kept on standby for genuine scarcity events rather than daily cycling. That’s an important distinction from the way gas used to operate. The plant isn’t disappearing. Its utilisation rate is.
Batteries, meanwhile, are being built at a scale that would have seemed implausible five years ago, but duration is still the constraint everyone in the industry quietly worries about. I wrote a while back about the data showing the median storage duration of the existing battery fleet sitting closer to two hours than four, which matters enormously for whether these assets can actually do the multi-hour evening job the market increasingly needs them for, not just the sub-hour frequency work they were originally built around: a point covered in Are big batteries being built for the wrong job?
My honestly mildly contrarian view #
If I’m putting a stake in the ground: I reckon the industry oversold gas peakers as a transition bridge for longer than the economics justified, and it’s now overselling batteries as a full substitute for longer-duration firming before the fleet actually has the duration to do it. Both of those are marketing problems more than engineering ones. The engineering answer is fairly boring. You need short-duration batteries for the daily ramp, longer-duration storage (pumped hydro, four-hour-plus batteries, or gas) for the extended gaps, and you need the market signals to reward each for the specific job it does rather than paying them as if they’re substitutes.
The bit that gets lost in the public argument is that gas and batteries aren’t really competing for the same contract most of the time. They’re competing for adjacent slices of the same problem, and the overlap is smaller than the headlines suggest. Where they do genuinely compete, in the four-to-eight-hour band, batteries are closing the gap fast, helped by falling battery costs and the CEFC’s continued co-investment in storage projects, but they haven’t closed it completely, and anyone telling you gas peakers are finished in the NEM within this decade hasn’t looked closely at the AEMO reliability forecasts for the regions reliant on a single interconnector, like Tasmania, which we’ve covered in Tasmania’s Battery of the Nation bet still rests on one cable.
So which one wins? #
For the daily ramp, the job that happens every single evening across the NEM, batteries win, and increasingly they win by a wide margin on cost per dispatched megawatt-hour. For genuine multi-day scarcity events, gas still wins, because nobody has yet built battery fleets with the duration or the fuel-agnostic flexibility to replace it, and the economics of building batteries purely for rare events don’t stack up without very specific capacity payments.
The honest scorecard, if you want one number: on a typical summer evening peak in 2026, batteries are already doing more of the heavy lifting than gas in states like South Australia and Victoria. On a cold, still week in July with low wind output across three states at once, gas still matters more than most renewables advocates want to admit. Both of those statements are true at the same time, and the AEMC’s ongoing capacity mechanism design work is essentially an attempt to pay for both roles properly rather than pretending one technology has made the other redundant.
Anyway – back to that drawn endgame in Brunswick. I resigned eventually, not because the position was lost but because I’d mistaken which piece needed to do which job three moves earlier. The NEM’s firming debate has the same lesson sitting right there if people would just look at it.
– Anjali Rao, Grid & Storage Correspondent
Photo by Jonathan Hanna on Unsplash