Two hours. That’s the average discharge duration of most big batteries now under construction or contracted across the National Electricity Market, and it’s the number nobody spruiking the battery boom wants to sit with for very long. Follow the money on any of these projects and you find developers chasing frequency control ancillary services and price arbitrage in the five-minute market, not the multi-day firming job that the public conversation keeps assuming batteries are built to do. Those are two different jobs. Only one of them pays the bills right now.
The duration problem nobody prices in #
Australia’s big batteries have got bigger, fast. The projects announced or under construction over the past two years, from Akaysha Energy’s build-out to the Waratah Super Battery near Munmorah in New South Wales, run into the hundreds of megawatts. That’s real, and it matters for grid stability.
But megawatts aren’t the whole story. Duration is.
A battery rated at 300 MW with two hours of storage delivers 600 MWh before it’s empty. That’s roughly enough to cover an afternoon peak, or to smooth the ramp when the sun drops and rooftop solar output falls off a cliff. It is nowhere near enough to get a grid through a multi-day wind lull in winter, the kind south-eastern Australia sees most years. AEMO’s own Integrated System Plan is explicit that the grid needs deep, long-duration storage as coal retires — think days, not hours. Two-hour batteries were never designed to solve that problem, and mostly nobody claims they were. The trouble is the public debate has started treating “big battery” and “grid firming” as interchangeable, and they aren’t.
What batteries are actually paid to do #
Batteries earn their keep in the fast markets. Frequency control ancillary services, arbitrage between the cheap midday hours when rooftop solar floods the grid and the expensive evening peak, and increasingly a bit of both stacked together. That’s the commercial reality, and it’s a sensible one — it’s also why negative electricity prices at midday have become such a fixture of the NEM’s daily rhythm. Batteries charge on the negative-price hours and discharge into the evening ramp. It’s a clean trade, and it’s the one the market pays for.
The AEMC and AER have both noted, in various market design reviews, that the current settlement structure rewards short-duration flexibility handsomely because five-minute dispatch and the FCAS markets are where the volatility lives. For a primer on how that dispatch mechanism actually works, our explainer on how the NEM dispatches power every five minutes is worth a read. The point for today is simpler: developers build for the revenue stack that exists, not the reliability gap that policy papers describe. That’s not a criticism of the developers. It’s just how capital behaves.
Follow the money: arbitrage over firming #
The honest read is that most of the current battery pipeline is optimised for the trade that pays today, and the trade that pays today is short-duration arbitrage plus FCAS, not multi-day firming. Why would a rational investor build four hours of duration, at meaningfully higher capital cost per megawatt, when two hours captures almost all of the available revenue in the current price shape? They wouldn’t, and largely they haven’t.
This isn’t a scandal. It’s a market doing exactly what it’s designed to do. But it does mean the reliability story being told about batteries — that they’re quietly solving the coal-exit firming problem — is, at minimum, incomplete. Our piece on why wholesale electricity prices swing so violently covers the mechanics of why that volatility exists in the first place, and it’s exactly the volatility that two-hour batteries are built to milk.
The Capacity Investment Scheme’s blind spot #
The federal government’s Capacity Investment Scheme has poured contracted capacity into storage and renewables at a pace that’s genuinely reshaped the investment pipeline. We’ve asked before whether the scheme is quietly picking winners, and duration is part of that question. The CIS tender design does allow for longer-duration storage, and the scheme has broadened its scope over successive rounds. But the bulk of what’s actually been built and connected to date skews short. If the CIS is meant to plug the reliability gap left by closing coal, as our analysis of whether Australia is closing coal faster than it can replace it lays out, then duration mix matters as much as headline megawatts. A gigawatt of two-hour batteries and a gigawatt of eight-hour pumped hydro are not the same asset, even though a press release can make them sound identical.
Pumped hydro’s slow-motion comeback #
Which brings us to the asset that’s actually built for the multi-day job: pumped hydro. Snowy 2.0 has become the poster child for how badly these projects can run over time and budget, and that history is fair to raise. But the physics haven’t changed just because the delivery has been slow. Pumped hydro can store energy for days rather than hours, and it’s the natural complement to a battery fleet built for speed rather than depth. We’ve laid out the trade-offs in detail in pumped hydro vs big batteries: which firms the grid?, and the short version is that Australia probably needs both, in different proportions to what the current pipeline suggests.
The mildly contrarian bit, and I’ll own this one: I think the industry’s enthusiasm for batteries has let policymakers avoid the harder conversation about long-duration storage and transmission, because batteries are quicker to build, easier to finance, and produce better headlines. That’s a reasonable reason to build batteries. It’s a poor reason to declare the firming problem solved.
The wind drought test #
South Australia is the best live laboratory for this. The state runs on wind and solar for most hours of most days, and its record — covered in our piece on South Australia’s renewable energy record — shows both how far a grid can lean on renewables and where the gaps still show up. When wind output falls across the whole southern half of the NEM for two or three days running, which happens most winters, a two-hour battery fleet discharges once and then sits idle waiting for the next price spike. It’s gas, interconnection, or genuinely long-duration storage that carries the grid through that stretch, not the battery fleet everyone points to on the six o’clock news. AEMO’s reliability forecasts have flagged this gap directly, and it’s worth reading the regulator’s own published market data rather than taking anyone’s press release at face value on this one — the AER’s market performance reporting is the more sober source.
What the right job looks like #
None of this is an argument against batteries. It’s an argument for calling the job what it is. Batteries are excellent at absorbing midday solar, smoothing the evening ramp, and holding frequency steady in a grid that’s lost the physical inertia coal turbines used to provide for free. That’s a genuinely valuable job, and it’s one batteries do better and cheaper than almost anything else available. It is not, on its own, a substitute for firming through extended low-renewable stretches, and pretending otherwise sets up the grid for an uncomfortable reliability surprise sometime in the next few winters.
The consumer-facing side of this matters too — households weighing up rooftop solar against a home battery, as we covered in rooftop solar versus a home battery, are making the same duration trade-off on a smaller scale, and the economics there are just as duration-sensitive.
Somewhere in a planning meeting for the next CIS tender round, someone should be asking not “how many megawatts” but “how many hours, and for which job.” It’s about as gripping as watching a county side bat out a rain-shortened draw, I’ll grant you, but the boring questions are usually the ones that decide whether the lights stay on. The batteries being built right now are good at what they’re being asked to do. The honest read is that what they’re being asked to do, and what the grid actually needs done, have quietly drifted apart — and nobody’s fixed the invoice yet.
— Marcus Wren, Editor
Photo by Evgeniy Alyoshin on Unsplash