Six-fifteen on a still, hot evening in late January. Rooftop solar has fallen off a cliff, air conditioners across three states are pulling hard, and AEMO’s dispatch engine is hunting for the next few hundred megawatts in five-minute intervals. Somewhere a gas turbine at Tallawarra or Uranquinty is spinning up. Somewhere else, a battery the size of a football oval is already discharging, because it started half a second after the price signal arrived. Both did their job. They did not do the same job, and conflating the two is the single most common mistake in the firming debate right now.
I’ve sat through enough industry panels this year to lose count of the times someone says “batteries versus gas” as though it’s a simple substitution problem, like swapping one brand of turbine for another. It isn’t. One is a power problem. The other, mostly, is an energy problem. Let’s be careful with that distinction, because AEMO’s own Gas Statement of Opportunities treats it as central to how the market should be read.
The MW/MWh distinction everyone glosses over #
A gas peaker rated at 300 megawatts can, fuel supply permitting, hold that output for hours. It is constrained by gas availability and contract terms, not by an internal state of charge. A 300 MW battery is a different animal entirely — its megawatt rating tells you what it can deliver in an instant, but its usable megawatt-hours tell you for how long, and that number is usually two to four hours before it’s flat.
So when someone quotes a battery’s power rating as though it settles the firming question, my instinct is to ask: firms what, for how long? A battery sized at 300 MW / 600 MWh solves an evening peak beautifully. It does very little for a five-day wind drought in winter across South-East Australia, which is exactly the scenario AEMO’s Electricity Statement of Opportunities keeps flagging as the tail risk that keeps planners up at night.
What a gas peaker actually does at 6pm #
Open-cycle gas turbines are built for exactly this moment: cold start, full load inside ten to fifteen minutes, and the ability to keep running as long as gas keeps flowing. Snowy Hydro’s Kurri Kurri plant in the Hunter Valley — reported publicly in the 600–750 MW class depending on configuration — is the newest example in the NEM, and its business case rested on precisely this profile: short, sharp calls during evening peaks and supply gaps, not baseload running.
The catch is cost and carbon, not technical capability. Gas peakers burn expensive fuel exactly when the market is tightest, which is also when prices are highest, so their marginal cost tracks the spot gas price plus whatever carbon liability applies under the Safeguard Mechanism. Our piece on the Safeguard Mechanism goes into how that liability is now a real line item for gas generators, not an afterthought.
What a big battery actually does at 6pm #
A grid-scale battery does something a turbine physically cannot: it responds in under a second, and it can also absorb power, not just supply it. That’s a genuinely different service — frequency response, fast reserve, and increasingly the grid-forming inertia function that AEMO has been trying to procure more of, which I’ve written about in grid-forming inverters: why the grid suddenly needs them.
But the energy constraint is real and unforgiving. Most utility batteries built in the NEM over the past three years sit around a two-hour duration, some pushing toward four. That’s ample for an evening ramp. It is not ample for the multi-day wind lull that shows up in AEMO’s reliability modelling roughly once every few years. Batteries firm the shape of the day. Gas, hydro and interconnection firm the shape of the week.
A worked example worth sitting with #
Take a hypothetical but realistic evening peak: 300 MW of shortfall for four hours, from 5pm to 9pm.
- Gas peaker, 300 MW rated: delivers 300 MW × 4 hours = 1,200 MWh, limited only by gas supply and contract, with marginal cost driven by gas price plus carbon cost.
- Battery, 300 MW / 600 MWh: delivers 300 MW for two hours, then output must taper or the battery is empty — it covers 600 MWh of that 1,200 MWh requirement, half the task, before something else has to pick up the remaining two hours.
That’s not an argument against batteries — it’s an argument for pairing durations correctly with the shape of the shortfall. A four-hour battery closes that gap on its own. A two-hour battery needs a partner, whether that’s gas, pumped hydro, or imports across the interconnectors. Anyone telling you a single technology solves every hour of the year is selling something.
Who’s actually paying for this, and through what mechanism #
Most of the new firming capacity going into the ground right now, gas or battery, is being underwritten through the Commonwealth’s Capacity Investment Scheme, which effectively puts a floor and ceiling under revenue so projects can get financed against a genuinely volatile wholesale market. It’s worth remembering that scheme has skewed heavily toward batteries and renewables rather than gas, and I’ve argued elsewhere that the CIS is quietly picking winners rather than remaining technology-neutral, which changes the economics of new gas peaking more than most headlines admit.
Developers like Akaysha Energy, backed by BlackRock, have built a genuine pipeline of grid-scale batteries under CIS-style contracts — the strategy is covered in our profile of Akaysha Energy’s battery bet on the NEM. Meanwhile new gas peaking proposals face a tougher path to finance, partly because lenders increasingly price in the risk that a plant running a few hundred hours a year has a shrinking runway before it becomes a stranded asset.
Where I land, for now #
My honestly-not-that-controversial view: batteries have won the argument for daily peak-shaving, and it isn’t close. The economics, the speed of deployment, and the ancillary services stack up in their favour, and AEMO’s own dispatch data increasingly shows batteries setting price in the evening ramp rather than gas. But the multi-day, low-wind, low-solar tail risk that shows up in AEMO’s reliability standard modelling still needs something with genuine duration behind it — gas, pumped hydro such as the Kidston project profiled in our piece on Kidston, or firmed interconnection. Anyone who tells you gas peakers are finished, full stop, hasn’t sat with AEMO’s Electricity Statement of Opportunities long enough. Anyone who tells you we need to keep building gas at the old rate hasn’t looked at how fast battery costs and deployment timelines have moved either.
I’ll admit I used to think the answer was simply “more of both, don’t overthink it.” I’ve come around to a more specific view: the market needs batteries for the daily shape and something with real multi-day duration for the tail, and right now gas is still cheaper to build for that tail than long-duration storage, even if it’s not cheaper to run. That’s an uncomfortable answer for people who want a clean villain in this story. There isn’t one — there’s a duration mismatch, and the market is still working out how to price it properly.
There’s a chess parallel I keep coming back to, and I’ll make it quickly because I know this readership didn’t come here for openings theory. In an endgame with limited material, the player who wins isn’t the one with the most pieces on the board — it’s the one whose pieces cover the squares that actually matter. Firming the NEM is the same problem. It’s not about raw megawatts or raw megawatt-hours in isolation. It’s about which asset covers which hour, and right now nobody owns the whole board.
— Anjali Rao, Grid & Storage Correspondent
Photo by Alexandru Boicu on Unsplash