I spent a wet Tuesday last month at a community consultation session in the Hunter, where a council planner asked the developer of a proposed battery project a question I’ve heard a dozen times in different accents: “why not just build the gas plant, it’s cheaper.” The developer didn’t have a thirty-second answer. Neither do I, honestly, and that’s really the point of this piece. The comparison between gas peakers and big batteries for firming gets flattened into a single line in too many public debates, when it’s actually three or four separate questions stacked on top of each other.
Let’s be careful with that number before we go further, because “cheaper” means almost nothing without a unit attached. A gas peaker and a battery are not substitutes in the way a diesel car and a petrol car are substitutes. They do different jobs across different time horizons, and AEMO’s own Integrated System Plan treats them as complementary rather than competing technologies for most of the forecast period. That nuance tends to get lost the moment someone puts a dollar-per-megawatt figure on a slide.
What each technology is actually built to do #
A gas peaker, in the NEM context, is usually an open-cycle gas turbine. It can start from cold and reach full output in minutes, it burns gas the whole time it runs, and its economics depend on how often it’s actually called, which for a true peaker might be a few hundred hours a year. A big battery, the kind Neoen or Akaysha build, delivers power instantly from a standing start because there’s no combustion cycle to spin up, but it is fundamentally energy-limited. A 300 MW battery with two hours of storage holds 600 MWh. Run it at full power and it’s empty in two hours, full stop. There is no “topping up” mid-dispatch the way a gas plant can keep burning fuel for as long as the gas is flowing.
That distinction between power and energy is the whole ballgame here, and I think it’s where most kitchen-table arguments about this topic go wrong. Batteries are extraordinary at absorbing a sudden swing: a cloud bank rolling over a solar farm, a unit tripping offline, the five-minute dispatch gap AEMO has to fill every single interval, which I wrote about in a separate piece on how the NEM dispatches power every five minutes. Gas peakers are better suited to the slower, longer gap: a still, overcast week in winter when wind and solar output sag for days, not minutes.
A worked example, because round numbers hide the truth #
Take a 250 MW gap that needs filling for four hours during an evening peak. A battery sized at 250 MW / 500 MWh (two-hour duration, which is roughly what’s been getting built across the grid lately) covers two of those four hours at full output, then it’s done – unless it was charged earlier in the day and has more energy in reserve, which assumes cheap solar was available to charge it, which assumes the sun was out, which on a still overcast evening it may not have been. A 250 MW open-cycle gas turbine covers all four hours as long as gas supply and contracts hold up, constrained mainly by fuel cost and, increasingly, by fuel availability on the east coast: the subject I’ve gone over in detail when covering Santos and the east-coast gas standoff.
Now stretch the battery to four hours of duration: 250 MW / 1,000 MWh, and the comparison changes again, because four-hour batteries cost meaningfully more per megawatt-hour of installed capacity than two-hour ones, even though the marginal cost of running them once built is close to zero. The gas plant’s capital cost is lower per megawatt, but every hour it runs it burns fuel priced against a gas market that’s been anything but calm.
| Attribute | Gas peaker (OCGT) | Big battery (BESS) |
|---|---|---|
| Start-up to full output | Minutes | Sub-second to seconds |
| Duration limit | Fuel-supply limited, effectively unlimited | Energy-limited, typically 1-4 hours |
| Marginal running cost | Gas price plus carbon cost | Near zero once charged |
| Capital cost per MW | Lower | Higher, falling each year |
| Best suited task | Multi-hour or multi-day energy gaps | Fast frequency response, short peaks |
Where the Capacity Investment Scheme is actually pointing money #
The federal government’s Capacity Investment Scheme, which I broke down properly in a piece on how the Capacity Investment Scheme actually works, has leaned hard toward storage in its tender results so far, and that’s not an accident of politics. It reflects a judgement that the grid’s most acute near-term problem is the fast, frequent, short-duration gap, not the rare multi-day drought of wind and sun. Batteries are winning that argument on cost trajectory alone: the Clean Energy Regulator’s data and various state tender results both show battery costs falling faster than almost any other generation or storage technology tracked in Australia.
But the CIS has left room for gas too, and projects like Waratah Super Battery in New South Wales, which I covered when it started earning its keep as a literal shock absorber for the grid (see Waratah Super Battery: the shock absorber finally earning its keep), sit alongside new peaking gas capacity still being proposed in Victoria and Queensland. Nobody serious is arguing for an all-battery or all-gas firming fleet. The argument is about the ratio, and about which one gets built where.
The emissions and fuel-security angle nobody wants to say out loud #
A gas peaker running three hundred hours a year has a genuinely small annual emissions footprint compared to a baseload coal unit, and that’s the honest case gas proponents make, one I think deserves more airtime than it gets in some renewables-only commentary. The trouble is fuel security, not emissions intensity per MWh. Victoria and New South Wales both depend on gas fields and pipeline capacity that are under real strain, a topic covered at length in pieces on the east-coast gas question generally. A peaker is only as reliable as the gas behind it, and on the coldest nights of winter, exactly when peakers are needed most, residential heating demand competes directly with generators for the same molecules.
Batteries don’t have that problem, obviously, but they introduce a different one: they need to have been charged by something, and if that something was a coal unit running overnight because wind was flat, the emissions accounting gets murkier than headline figures suggest.
Cost is moving, and the gap is closing faster than either side’s talking points admit #
I’d flag one mildly unpopular view here. The debate online often treats battery costs as static and gas costs as rising, which supports whichever conclusion the commentator wanted anyway. In reality both are moving targets. Battery cell costs have fallen for years on global manufacturing scale, largely driven by electric vehicle demand rather than anything happening in Australia, and that tailwind should keep helping here. Gas turbine costs haven’t fallen in the same way, but gas peakers built a decade ago are still running fine, and the capital is already sunk, which matters when you’re comparing a new battery build against an existing peaker rather than a new one.
The fairest comparison, in my view, is new-build against new-build. And on that basis, for anything under about three hours of duration, batteries are now routinely winning tender processes on a pure levelised cost basis, a trend the AEMC and AER’s published market data both support without much ambiguity. Beyond four hours, it gets genuinely contestable, and that’s where pumped hydro also enters the conversation, which I’ve set out properly in pumped hydro versus batteries: what actually firms the grid.
What this looks like on the ground, not the spreadsheet #
I rode my bike past a gas peaker site near Newport a few weeks back, the kind of unglamorous industrial shed most people never notice unless they’re specifically looking for it. No cooling towers, no plume, just transformers and a turbine hall. It’s been there decades, it runs rarely, and when it does run it earns a premium that looks absurd on a per-MWh basis until you remember it might only generate for forty hours in a whole year. That’s the economics of scarcity value, and it applies to batteries too, just compressed into a much shorter and much more frequent window. The grid needs both kinds of scarcity covered, and pretending one technology makes the other obsolete is the quickest way to get the firming debate wrong.
My actual verdict #
For the fast, frequent, short-duration gap (the five-minute dispatch interval, the cloud shadow, the sudden trip) batteries are now the better tool and the cost curve says that advantage will widen, not narrow, over the next few years. For the slow, rare, multi-day gap, particularly in a cold southern winter when gas demand for heating and generation collide, a well-contracted gas peaker still earns its place in the portfolio, assuming the fuel supply behind it holds. Anyone promising that batteries alone solve firming, or that gas peakers are obsolete, is selling a simplified version of a genuinely two-dimensional problem. The sensible build-out, and broadly what AEMO’s own system plan assumes, is more batteries doing more of the daily work, with a shrinking but real gas peaking fleet held in reserve for the days batteries simply can’t cover. That’s not a tidy conclusion. It’s just the one the numbers actually support.
– Anjali Rao, Grid & Storage Correspondent
Image: Maksym Kozlenko (CC BY-SA 3.0) via Wikimedia Commons