288. That’s how many times a day AEMO’s dispatch engine re-solves the entire National Electricity Market, picks a price, and tells generators what to do. Not once an hour. Not once a half-hour, which is how it used to work until 2021. Every five minutes, around the clock, for every one of the five mainland regions. Most people who pay an electricity bill have no idea this clockwork exists, and most who do think of it as a technical footnote rather than the actual mechanism that decides who gets paid and who gets switched off.
I’ve sat through a couple of AEMC technical briefings on dispatch design over the years, the kind where half the room is engineers and the other half is traders doing mental arithmetic on marginal cost curves. It’s a genuinely elegant system. It’s also the thing most commentary on “the market” gets wrong, because people talk about electricity prices as though they’re set like a supermarket shelf price. They’re not. They’re re-auctioned, nationally, every 288th of a day.
The five-minute settlement shift that changed the incentive #
Until October 2021, the NEM ran on five-minute dispatch intervals but settled prices on a 30-minute average of six of those intervals. That gap mattered more than it sounds. A battery or a fast-start gas peaker could spike output for one or two five-minute intervals, catch a high price, then coast through the rest of the half hour, and still get paid the averaged 30-minute rate. It rewarded speed without quite paying for it properly.
Five-minute settlement, which the AEMC pushed through as a rule change years in the making, fixed that mismatch. Now what you’re paid is what happens in that five-minute window, full stop. It’s a big part of why big batteries have become the preferred weapon for firming rather than mid-sized peaking gas. The economics finally reward the asset that can react in under a second rather than the one that can only react inside half an hour. For a fuller account of how that race between batteries and gas peakers is actually playing out on cost, I’d point to our earlier look at gas peakers versus big batteries and the firming maths in 2026.
What actually happens inside one dispatch interval #
Every generator registered in the NEM submits a bid stack, essentially a price-quantity schedule showing how much it’s willing to generate at each price band, up to ten bands per unit. Every five minutes, AEMO’s dispatch engine takes every one of those bids across the entire grid, adds in the forecast demand for each region, checks transmission constraints between regions, and solves for the cheapest combination of generation that meets demand without breaching any network limit. That’s not an estimate. It’s a formal linear-programming optimisation, run against real engineering constraints on interconnectors and local networks, with a result spat out in well under the five minutes it covers.
The price that clears that interval, the dispatch price, becomes the basis for the spot price used in settlement. Whoever bid at or below that clearing price gets dispatched and paid the clearing price, not their own bid, which is the standard marginal-pricing logic used in most well-functioning electricity markets worldwide. It’s also why bidding strategy in the NEM is a specialist discipline in its own right. A generator that bids too high risks being skipped over entirely for that interval. Bid too low and you leave money on the table if the market clears higher than your offer.
Why constraints matter as much as the bid stack #
Here’s where dispatch gets genuinely interesting, and where a lot of outside commentary falls down. The cheapest generation on paper isn’t always the generation that gets dispatched, because the transmission network has physical limits. If a line between regions is constrained, AEMO’s engine has to respect that even if it means dispatching a more expensive generator closer to load instead of a cheaper one stuck behind a bottleneck.
This is the mechanical reason interconnector projects like VNI West and Marinus Link matter so much to the economics of the whole system, not just to the regions they directly connect. A bottleneck anywhere in the network shows up as a price divergence somewhere else, five minutes at a time, every single day. Queensland and South Australia regularly diverge from the southern states purely because of constraint, not because the underlying generation mix is wildly different.
Renewable Energy Zones compound this. You can build gigawatts of wind and solar in a cheap-to-generate location, but if the poles and wires connecting that zone to load centres aren’t built or aren’t rated for the flow, dispatch has to curtail the cheap generation and call on something more expensive instead. That’s effectively what’s been happening in pockets of NSW, something I’ve gone through in more detail in our look at how NSW’s coal exit is outrunning its REZ build.
Where rooftop solar breaks the clean model #
The textbook version of dispatch assumes AEMO can see and control the generation fleet. Rooftop solar blows a hole straight through that assumption. There’s now something past 24 gigawatts of rooftop PV capacity sitting on Australian roofs, invisible to AEMO’s dispatch engine in the sense that it isn’t bid into the market at all. It just generates, whenever the sun is out, and net demand (what AEMO actually has to serve from scheduled and semi-scheduled plant) falls off a cliff around the middle of a sunny day.
AEMO doesn’t dispatch rooftop solar. It has to forecast it, subtract it from underlying demand, and then dispatch everything else against what’s left. When that forecast is wrong, which it sometimes is on patchy-cloud days, the whole dispatch solve for that interval is working off a slightly wrong number. This isn’t a hypothetical problem; it’s the daily reality AEMO manages, and it’s a big part of why minimum demand troughs have become such a live operational issue. We’ve covered the mechanics of that squarely in rooftop solar and the midday minimum demand problem.
The contrarian point I’d make here, and I don’t think it’s a popular one in rooftop solar circles, is that treating behind-the-meter PV as a free good with no market obligations is starting to look like a design choice the system can’t keep affording. At some point a meaningful slice of that fleet needs better visibility, and probably some form of dynamic export management, not because rooftop solar is the villain but because dispatch can’t optimise what it can’t see.
Grid-forming inverters and the physics dispatch doesn’t price #
There’s a layer underneath all of this that the five-minute price signal doesn’t directly capture: system strength and frequency stability. Traditional dispatch logic solves for cost and megawatts. It doesn’t, on its own, guarantee the grid has enough spinning inertia or fast frequency response to ride through a sudden fault without cascading.
That’s a separate but related market, the frequency control ancillary services market, running its own parallel five-minute (and faster) dispatch alongside the energy market. As coal units retire and more of the generation fleet becomes inverter-based rather than spinning machinery, AEMO has leaned harder on this ancillary layer, and on newer grid-forming inverter technology, to hold the system together. We’ve written separately about why that hardware distinction actually matters operationally in grid-forming inverters: why the grid suddenly needs them.
Why five-minute price swings aren’t a glitch #
People occasionally see a spot price chart with a spike to several thousand dollars a megawatt-hour for a couple of intervals, then back to double digits, and conclude the market is broken. It isn’t. That’s dispatch doing exactly what it’s designed to do: signal scarcity, for exactly as long as the scarcity lasts, no longer.
A short, sharp price spike during a hot afternoon peak, followed by a rapid collapse once a battery or a peaker responds, is the system working. The price cap on the NEM, set and periodically reviewed by the AEMC, exists precisely to stop that signal running away entirely while still letting it bite hard enough to bring fast capacity online. I’ve written before about why these swings look so dramatic and why that’s mostly the point, in our explainer on why wholesale electricity prices swing so violently.
What this means for the next five years #
The mechanism itself isn’t going to change much. What’s going to change is the mix of things being dispatched into it. More batteries bidding tighter, faster responses. More grid-scale solar and wind, each needing careful treatment because their output is a weather forecast as much as a bid. Gas plant increasingly used for genuine peak scarcity rather than baseload, which changes how operators like those running the assets discussed in our profile of Kurri Kurri actually have to think about bidding strategy.
AEMO’s own published dispatch data, which anyone can pull from the market data portal, tells this story better than any press release. Five-minute dispatch isn’t glamorous. It’s the quiet machinery underneath everything else in this transition, and it’ll keep re-running its sums every 288th of the day whether anyone’s watching or not. Worth remembering that the next time a bill arrives and the instinct is to blame “the market” as some vague abstraction, rather than a very specific optimisation problem being solved, again, right now, somewhere in a control room in Homebush.
– Marcus Wren, Editor
Photo by Compagnons on Unsplash