Twenty-four gigawatts. That’s roughly where Australia’s rooftop solar fleet sits now, spread across more than four million homes and small businesses, according to the Clean Energy Regulator’s small-scale scheme data. It is bigger than the entire coal fleet still running in the NEM. And on a mild, sunny Sunday in South Australia, it can supply more electricity than the state needs, pushing operational demand into negative territory before lunch. That is not a hypothetical. It has happened repeatedly, and AEMO has been publishing the numbers for years.
So the question in the title isn’t really a stretch. Rooftop solar grid stability has become a genuine technical problem for the system operator, not just a talking point for people who don’t like solar. The honest read is that the panels themselves aren’t the issue. What’s actually straining the grid is a market and tariff structure built for one-way power flows, trying to cope with millions of small generators nobody centrally controls.
What “breaking the grid” actually means here #
Nobody serious is suggesting rooftop solar is going to cause statewide blackouts on its own. The risk is narrower and more technical than that. It’s about minimum demand.
The grid needs a certain amount of load being met by large, spinning generators to hold frequency and voltage steady. When rooftop solar pushes underlying demand down towards zero, or below it, there’s less room for those big machines to do their job, and less system strength to fall back on if something trips. AEMO’s engineering framework work has flagged minimum demand as one of the fastest-moving risks in the whole transition, arguably faster moving than the more talked-about peak demand problem. We wrote about the mechanics of this in negative electricity prices at midday: what they mean for you, and the pattern has only sharpened since.
South Australia is the pointy end of it, which is why we keep coming back to that state. Its solar penetration is high enough that on some days rooftop PV alone can technically meet, or nearly meet, the entire state’s demand. We went through what that record actually hides in South Australia’s renewable energy record: what it hides, and minimum demand is the chapter that gets skipped in the celebratory press releases.
The backstop nobody asked for #
South Australia was first to act, requiring new solar inverters to carry remote disconnection capability under reforms that date back to 2020. It wasn’t popular. But it meant that when minimum demand risk spiked, the network could dial output down rather than face an uncontrolled system event. Victoria has since built its own emergency backstop mechanism, giving distribution businesses the ability to curtail rooftop exports during periods AEMO flags as high risk.
Follow the money on this one and it gets interesting. The cost of building that remote curtailment capability, the metering, the communications infrastructure, the compliance overhead, sits mostly with distribution businesses and, eventually, with all customers through network charges. Households that installed solar under the old rules, with an uncapped, always-on export arrangement, are not the ones paying for the fix. Everyone else is, through the poles-and-wires component of their bill. That’s a distributional question worth more scrutiny than it gets, and it’s one we’ve poked at before in Who really pays for solar?
Flexible exports were meant to solve this #
The AEMC ran a rule change back in 2021 designed to fix exactly this problem before it became acute: flexible export limits, where a household’s solar export cap moves with real-time network conditions instead of being fixed at a flat, conservative number. In theory, a smart inverter gets told to export 1.5 kilowatts at 1pm on a low-demand Tuesday and 10 kilowatts on a windy, high-demand afternoon, instead of the network defaulting everyone to the lowest common denominator to protect against the worst case.
In practice, rollout has been patchy. Distribution businesses vary enormously in how far they’ve implemented dynamic operating envelopes, and the software, metering and retailer coordination needed to make it work properly is still catching up in a lot of regions. Energy Consumers Australia and the state distribution businesses have both flagged the gap between the rule change and what’s actually live on the ground. That lag is the real story, not the solar panels.
The tariff problem underneath the technical one #
Here’s my mildly contrarian bit, and I’ll own it. The consensus framing, that Australia has “too much” rooftop solar and needs to slow installations down, is the wrong diagnosis.
The actual problem is that retail and network pricing hasn’t caught up with what solar does to the demand curve. Feed-in tariffs in most states have collapsed towards zero, or gone negative at the times solar floods the grid, which we mapped out properly in how rooftop solar feed-in tariffs actually work in 2026. That price signal should, in theory, tell new solar owners to add a battery or shift consumption. Instead, most households still get billed on flat or lightly time-varying tariffs that don’t reflect the five-minute reality of the grid at all. We’ve argued for cost-reflective pricing reform in Air-conditioners and solar — why electricity pricing needs to be reformed, and minimum demand is exactly the kind of problem that pricing reform, done properly, would help solve without anyone having to remotely switch off a neighbour’s inverter.
Blaming the panels is the easy version of this story. It lets policymakers avoid the harder conversation about network tariff reform, and it lets retailers avoid explaining why feed-in tariffs fell off a cliff while retail prices mostly didn’t. I reckon that’s not an accident.
Batteries, EVs and the slow fix #
The Consumer Energy Resources roadmap that state and federal governments have been working through gives a fair sense of where this is heading, and we covered the practical detail of it in Australia’s Consumer Energy Resources roadmap: what it means on the ground. Home batteries soak up the midday glut and release it in the evening peak, which is the single most useful thing a household can do for grid stability right now, arguably more useful than the panels themselves at this point in the cycle. We ran the actual payback numbers in rooftop solar versus a home battery: which pays back first, and the economics are shifting fast enough that this stops being a niche argument within a couple of years.
EVs charging overnight or on smart schedules do something similar on the demand side. None of it happens quickly enough to solve the minimum demand problem this coming summer, though. AEMO’s own quarterly dynamics reporting has flagged minimum demand risk periods getting earlier and deeper each year, and the system strength projects meant to backstop that, the synchronous condensers and grid-forming batteries we covered in grid-forming inverters: why the grid suddenly needs them, are still being built out region by region.
What I’d actually watch #
I spent an afternoon a few months back at a distribution control room in Adelaide, watching an operator flick between live export data and a weather radar, trying to picture what a cloud passing over a suburb does to voltage on a low-voltage feeder three streets away. It’s genuinely fiddly work, closer to wicketkeeping standing up to the stumps than the neat spreadsheet version of the grid people imagine from the outside. You’re reacting to something small and fast, over and over, and mostly nobody notices when you get it right.
The number to watch isn’t installed rooftop capacity. It’s the proportion of that fleet sitting on dynamic export limits versus fixed ones, and how fast distribution businesses actually finish the flexible exports rollout the AEMC mandated years ago. That’s the honest scoreboard. Everything else, the minimum demand headlines, the emergency backstop stories, the periodic panic about “too much solar,” is downstream of that one unglamorous infrastructure project either getting finished or not.
Rooftop solar didn’t break the grid. It exposed a pricing and control system that was never built to handle four million independent generators, and we’re now paying, quite literally through network charges, to retrofit it. Whether that bill lands fairly is the argument actually worth having.
— Marcus Wren, Editor