Two hundred and eighty-eight times a day, in every region of the National Electricity Market, a piece of software called NEMDE runs an auction that most Australians have never heard of and that quietly sets the price of almost everything downstream — your power bill, a battery developer’s revenue model, whether Loy Yang runs flat out or backs off. That’s the five-minute dispatch cycle. It’s the engine room of the NEM, and it’s a lot simpler in concept than the jargon around it suggests.
The auction nobody watches #
Every five minutes, generators across Queensland, New South Wales, Victoria, South Australia and Tasmania submit bids to the Australian Energy Market Operator. Not a price for the whole day — a price for that specific five-minute slice, and the next one, and the one after that, out to a rolling forecast.
A coal unit might offer its output at $40 a megawatt-hour. A gas peaker might sit at $300. A wind farm might bid at zero, or even below zero, because it’s chasing renewable certificates rather than the spot price itself. AEMO’s dispatch engine, NEMDE, stacks every bid from cheapest to most expensive and works down that stack until forecast demand is met. Whatever the last, most expensive generator needed to meet demand costs, that becomes the price paid to every generator dispatched in that interval — the marginal price. It’s not complicated maths. It’s just done relentlessly, all day, every day, for every region.
Why five minutes and not thirty #
The NEM didn’t always work this way. For its first two decades the market settled on the old six-dispatch-interval average — physical dispatch happened every five minutes but the price consumers and generators actually settled against was smoothed over thirty-minute blocks. That mismatch became a genuine problem once fast-acting batteries and gas peakers started gaming the gap between dispatch and settlement, chasing the thirty-minute average rather than reacting to the real five-minute signal.
The Australian Energy Market Commission ruled in 2017 to close that gap, and five-minute settlement finally went live in October 2021 after a long and, frankly, bruising implementation fight with incumbent generators who preferred the old system. The honest read is that the change was overdue. Fast response assets — batteries especially — now get paid for exactly what they deliver in the interval they deliver it, not for gaming an average. Follow the money and you can see why some incumbents fought it: five-minute settlement rewards flexibility, and coal and old-style gas plant don’t do flexibility well.
What NEMDE actually optimises #
Dispatch isn’t just about cheapest megawatt in, cheapest megawatt served. NEMDE co-optimises energy with frequency control ancillary services — the eight FCAS markets that keep the grid’s frequency at 50 hertz. A battery bidding into the market is often weighing up whether it earns more supplying raw energy or holding capacity back for a fast-frequency-response contract. That’s part of why Akaysha Energy’s big batteries and others like them don’t simply charge low and discharge high — they’re running a constant, five-minute-by-five-minute calculation across multiple markets at once.
Network constraints get bolted on too. If a transmission line between two regions is running near its thermal limit, NEMDE has to respect that even if a cheaper generator sits on the wrong side of the bottleneck. This is where dispatch gets genuinely messy — and, in my view, it’s the part of the system getting quietly more important than the headline spot price. As more wind and solar clusters in regions like western Victoria and central-west New South Wales, the binding constraint on any given afternoon is increasingly the wire, not the generator. AEMO’s own integrated system plan flags transmission as the long pole in the tent, and dispatch data bears that out most days you look at it.
Why the price jumps around so much #
This is the bit that confuses people who check their app once a week and see the spot price at $15,000 a megawatt-hour on a Tuesday evening in February. Because price is set by the marginal generator in each five-minute window, and because demand and generation availability both move fast, the price can swing from negative to the market price cap within an hour.
We’ve covered why that volatility exists in more detail here — see why wholesale electricity prices swing so violently — but the short version tied to dispatch is this: there’s no smoothing left in the system anymore. Every five-minute price is real, is settled, and reflects exactly what it cost to serve that slice of demand. Midday troughs with heavy rooftop solar can push prices negative, sometimes for hours at a stretch, a phenomenon we’ve unpacked separately in negative electricity prices at midday: what they mean for you. Dispatch doesn’t average that away. It just reports it, honestly, every five minutes.
The battery advantage baked into the design #
Five-minute dispatch happens to suit batteries better than almost any other asset class in the NEM, and that’s not an accident — it’s a direct consequence of the 2021 settlement reform. A battery can respond in seconds. A gas peaker takes minutes to synchronise. A coal unit can take hours to ramp meaningfully. When the reward structure is genuinely five-minute granular, the fastest asset wins the arbitrage most consistently.
That’s a big part of why the Capacity Investment Scheme — which we’ve examined in is the Capacity Investment Scheme quietly picking winners? — has leaned so heavily toward batteries and firmed renewables rather than new gas. The dispatch mechanics reward speed, and speed is what batteries sell. Pumped hydro plays a different, slower game in this same market, discharging over hours rather than minutes, and we’ve set out that trade-off properly in pumped hydro vs big batteries: which firms the grid?
Where the system still creaks #
None of this means dispatch is a solved problem. The constraint equations that govern which generator can physically deliver into a congested part of the grid are getting more numerous and more binding as renewable generation clusters in a handful of regions. AEMO publishes these constraint sets under the National Electricity Rules, and reading through a typical day’s dispatch data now involves wading through dozens of active network constraints rather than the handful that used to apply a decade ago.
There’s also a genuine debate, not fully resolved, about whether five-minute dispatch prices are becoming too noisy a signal for long-term investment decisions — a battery developer can read five years of five-minute data and still get blindsided by a single new constraint on a transmission corridor. The Australian Energy Regulator’s own market monitoring has noted the growing complexity of constraint management as more variable renewable generation enters the system. I reckon the market design itself is sound; the transmission build to keep pace with it is the part running behind, roughly the way a middle order collapses not because the top order failed but because nobody built the platform for them to bat on.
What this means for the next stage of the transition #
The five-minute market is, on balance, doing its job. It rewards flexibility, punishes slow-responding plant, and gives batteries and firmed renewables a clean commercial signal to chase. That’s not a small thing — a lot of the investment case for projects like those covered in our state renewable energy targets compared piece rests on dispatch mechanics working exactly as designed.
Where I’d push back on the consensus is the assumption that dispatch price volatility is itself the problem needing fixing. It isn’t. It’s the accurate symptom of a grid short on flexible, dispatchable capacity in the wrong places at the wrong times. Treat the symptom and you get gas subsidies and blunt price caps. Treat the cause — more storage, more transmission, better constraint management — and the five-minute price starts looking calmer on its own. AEMO’s National Electricity Market data dashboard is public and free to watch; spend an afternoon on it during a heatwave and the whole system starts making a lot more sense than any explainer, including this one, can manage in fifteen hundred words.
— Marcus Wren, Editor
Photo by Andrey Metelev on Unsplash