Two hours. That’s the median storage duration AEMO’s own battery fleet data has hovered around for most of the last three years, even as the headline megawatt numbers keep climbing. Everyone quotes the capacity figure when a new battery gets announced. Almost nobody quotes the duration. That gap is the whole story.
Australia now has gigawatts of grid-scale battery capacity either running or under construction, and the investment case for nearly all of it was built around price arbitrage and frequency control, not keeping the lights on through a still, cold evening in July. That’s a very different job to the one most people think batteries are doing. And it matters, because the public conversation about batteries has quietly drifted into assuming they’re the long-duration firming answer to coal retirements, when the commercial incentives driving most of the build tell a different story.
What the money is actually chasing #
Start with how a battery developer actually gets paid. The bulk of revenue for most big batteries in the NEM comes from two places: arbitrage between cheap midday solar and expensive evening peak prices, and participation in the frequency control ancillary services market, FCAS. Both reward speed and short bursts, not duration. A battery that can charge at 1pm and discharge hard for ninety minutes at 6pm, then flip into FCAS the rest of the time, is doing exactly what its spreadsheet told it to do.
Look at the economics piece I wrote a while back on gas peakers versus big batteries and the firming maths – the same tension shows up there. Gas peakers get paid to sit idle for 350 days a year and then earn a fortune on the five days the market actually needs them. Batteries, built with two to four hours of storage, are structurally better suited to that exact kind of short, sharp dispatch than to riding out a multi-day wind lull. Nobody’s hiding this. It’s in every developer’s financial model and every AEMC market design paper on storage revenue stacks. The issue is what gets implied in the public debate versus what’s on the balance sheet.
The duration problem nobody wants to cost properly #
A battery sized for two hours of discharge is an excellent tool for smoothing the evening ramp when solar falls away and demand is still rising. It is a poor tool for getting a grid through a genuine wind drought across south-eastern Australia, the kind AEMO’s Integrated System Plan models running several days in a row under certain weather patterns. For that job you need storage measured in days, not hours – pumped hydro, or a much bigger and much more expensive battery fleet than anyone is currently proposing to build at scale.
This is where I think the industry commentary gets lazy. Every time a gigawatt-scale battery gets approved, there’s a round of coverage implying the reliability problem is basically solved. It isn’t. Projects like Kidston’s long slow walk to commissioning and Queensland’s Borumba bet exist precisely because batteries, as currently financed and sized, don’t cover that multi-day tail. Pumped hydro is slower to build and much more capital intensive upfront, which is exactly why fewer of them get built relative to batteries, but they’re solving a different reliability problem, one batteries aren’t currently priced to solve.
Waratah and the shift that’s actually happening #
To be fair to the sector, the duration conversation is moving. The Waratah Super Battery in the Illawarra was built explicitly as a grid stability asset, not an arbitrage play: its job is system strength and fault ride-through support for a grid losing synchronous coal generation, which is a genuinely different function to a merchant battery chasing the evening price spread. That’s a sign the market is starting to differentiate between battery types rather than treating all storage as fungible.
But Waratah is the exception that proves the rule. Most of the battery pipeline coming through state and federal tender processes, including projects shaped by the Capacity Investment Scheme’s tender settings, still gets contracted on capacity and dispatch flexibility rather than duration minimums that would force longer-duration designs. The CIS has nudged things toward four-hour systems in recent rounds, which is progress. It is not the eight-to-ten-hour benchmark that would actually close the multi-day reliability gap AEMO keeps flagging in its system strength and reliability reporting.
Where the AEMO numbers actually point #
AEMO’s 2026 Electricity Statement of Opportunities and the broader Integrated System Plan modelling both treat storage duration as a key sensitivity, not a rounding error. The ISP’s own scenario work shows that reliability outcomes degrade meaningfully once you model realistic correlated wind and solar droughts lasting more than a day or two, which is exactly the condition two-hour batteries can’t ride through no matter how many of them you stack side by side. You can read AEMO’s published Integrated System Plan documentation directly on AEMO’s website if you want the scenario detail rather than my summary of it.
The AER’s own reporting on contestable storage revenue, meanwhile, keeps confirming that arbitrage and FCAS dominate current battery earnings, which is the commercial signal telling developers to build short and fast rather than long and steady. You can find the AER’s wholesale market monitoring and storage revenue commentary through its published market reports. Two regulators, two different lenses, same conclusion: the physics says we need duration, the market signal says build for speed.
A cricket analogy, because it fits for once #
It’s a bit like loading your bowling attack with four quicks who can each bowl a fierce opening spell, and no one who can hold an end for twenty overs on a flat wicket. Brilliant in short bursts, exposed the moment the contest runs long. The NEM’s battery fleet right now is mostly quicks. Useful, dangerous even, but not built for the grind.
That’s not a knock on the developers. Akaysha, Neoen and the other big names building out the battery pipeline at Orana and Ulinda Park are responding rationally to the revenue stack in front of them. If the market pays best for speed, you build for speed. The contrarian point I’d make, and I don’t think it’s a popular one in the renewables commentary at the moment, is that the industry and some of the state energy departments have let the batteries-equal-firming narrative run further than the underlying asset mix actually supports. It’s not dishonest. It’s just imprecise in a way that matters once a coal unit actually trips offline in week two of a wind lull.
What actually needs to change #
None of this means batteries are the wrong technology. It means the current generation of grid-scale batteries is being asked, in the public imagination at least, to do a job its financing structure was never built for. Fixing that is partly a market design problem: longer minimum duration requirements in future CIS and state tender rounds, clearer separation in AEMO’s reliability standard between short-duration arbitrage capacity and genuine firming capacity, and partly a messaging problem. Every press release announcing a new battery project should say plainly how many hours of discharge it offers at full output, the way a car ad has to disclose fuel consumption. Most don’t, because two hours doesn’t sound as impressive as “300 megawatts.”
The states that get this right will be the ones that stop treating battery megawatts and pumped hydro megawatts as interchangeable line items in the same reliability ledger. Queensland’s energy plan, as I’ve written before looking at the pumped hydro pillar and the cliff face next to it, is at least explicit about needing both. Other jurisdictions are still letting the batteries do more rhetorical heavy lifting than their duration specs can actually carry.
The honest ledger #
So are big batteries being built for the wrong job? Not wrong, exactly. Mis-sold, maybe. They’re very good at the job the market currently pays them to do: smoothing volatility, propping up frequency, catching the evening ramp. They are not, as currently financed and sized, the long-duration reliability backstop the public debate keeps implying they are. Closing that gap needs either a lot more duration in the fleet, a lot more pumped hydro actually finished rather than announced, or an honest admission that gas peakers are going to be holding that line for longer than the transition timeline likes to admit.
Watch the next round of CIS contracts for duration minimums. That’s the number that will tell you whether the market design is catching up with what the grid actually needs, or whether we’re just going to keep building more quicks and hoping nobody notices there’s no one left to bowl the last session.
– Marcus Wren, Editor