Nineteen gigawatts of rooftop solar capacity sat on Australian roofs by the Clean Energy Regulator’s last count, and the figure keeps climbing by roughly three gigawatts a year. That’s not a rounding error on a grid built to run on coal and gas. It’s a second power system, owned by three million-odd households and businesses, operating with almost none of the obligations AEMO places on a wind farm or a gas peaker. So when someone asks whether rooftop solar is quietly breaking the grid, my first instinct is to push back on the framing. It’s not breaking anything by itself. It’s exposing a grid that was never designed for millions of uncoordinated, invisible generators pushing power the wrong way down the wires at lunchtime.
That distinction matters more than it sounds like it should.
The minimum demand problem is real and it’s getting worse #
I’ve written before about the midday minimum demand squeeze this masthead has tracked for several years now – see our earlier piece on rooftop solar and the midday minimum demand problem. The short mechanical fact hasn’t changed: on a mild, sunny weekend day, rooftop solar can supply well over half of operational demand in South Australia and push close to it in parts of Western Australia and Queensland. AEMO’s own Operational demand forecasting work for the NEM has flagged minimum demand troughs falling faster than almost anyone predicted a decade ago. When underlying demand drops that low, the handful of big synchronous machines still running have to work harder to hold frequency and voltage steady, because there are fewer of them left spinning.
South Australia is the live laboratory here, and it’s worth being blunt about what that state’s experience actually shows. On some days SA’s grid runs on rooftop solar alone for stretches, with large-scale generation and batteries mopping up the edges. I covered the uncomfortable parts of that record in a piece on South Australia’s renewable energy record – the bit that gets underplayed in the celebratory coverage is how much synchronous condenser and gas-peaker backup AEMO has had to procure specifically to keep that grid stable when rooftop output is high and demand is thin.
Export limits are the blunt instrument nobody voted for #
Here’s where the policy response gets messy. Distribution network businesses across the NEM have been quietly rolling out dynamic export limits and, in some feeders, static caps that cut a household’s ability to send solar back to the grid during the riskiest hours. Western Power in WA and several NSW and Queensland distributors have done versions of this. The logic is defensible from an engineering standpoint: a feeder built for one-way power flow genuinely cannot absorb unlimited reverse flow without voltage rise causing inverters to trip or, worse, nuisance outages for neighbours. But the politics are brutal. A household that paid for a 10-kilowatt system under one set of rules can find itself curtailed to 1.5 kilowatts export a few years later, with little warning and less compensation.
Our earlier explainer on the Consumer Energy Resources roadmap goes through the regulatory machinery AEMO and the AEMC are building to manage this more gracefully: dynamic operating envelopes, two-way tariffs, the lot. It’s sound engineering. It is also, I’d argue, a quiet admission that the original “set and forget” solar feed-in model from the 2010s was never going to survive contact with twenty gigawatts of capacity.
Where I think the consensus gets it backwards #
The standard line from a chunk of the renewables lobby is that rooftop solar is simply a victim of distributor under-investment, and that with enough grid-forming inverters and smarter tariffs the problem mostly disappears. I think that’s half right and half wishful. Grid-forming inverter technology, which I went through in detail in our explainer on why the grid suddenly needs them, genuinely helps with the synthetic inertia problem. But it doesn’t fix the deeper structural issue, which is that Australia built the world’s highest per-capita rooftop solar fleet with almost zero central coordination of *when* that capacity gets installed, on which feeders, facing which direction.
A wind farm goes through a formal connection agreement with AEMO, a generator performance standard, and years of modelling before it’s allowed to export a single megawatt. A rooftop system goes through a sparky, a Clean Energy Council accreditation check and a phone call to the retailer. Both end up injecting power into the same wires. Only one of them has to prove it won’t destabilise the local grid first.
That asymmetry is the actual story, not solar panels themselves. The panels are doing exactly what they were sold to do.
Batteries are the pressure valve, but the economics are still catching up #
The obvious fix is storage that soaks up midday solar and releases it in the evening peak, smoothing both ends of the duck curve. Household batteries are growing fast off a small base, and I’ve gone through the sums properly in home battery payback: does it stack up yet – the honest answer there was “almost, in the right state, with the right tariff.” The bigger lever is probably orchestrated batteries and virtual power plants aggregating thousands of small systems into something that behaves, from AEMO’s control room, a bit like a dispatchable generator. Our 2026 reality check on virtual power plants found the megawatt numbers still modest next to grid-scale batteries like the Waratah Super Battery, but the growth trajectory is the steepest thing on the board.
Here’s my mild contrarian view, and I’ll own it: I don’t think household batteries scale fast enough on current economics to solve the minimum demand problem this decade without heavier intervention – either mandated smart export control as standard, or a much more aggressive subsidy skew toward storage over panels. The Commonwealth’s small-scale battery rebate introduced this year is a start, but it’s playing catch-up against a solar fleet that’s already installed and already exporting. You can’t un-ring that bell. You can only manage around it.
What the grid operators are actually doing about it #
AEMO’s response, laid out across successive Integrated System Plan documents, treats distributed PV as a forecasting input rather than a fault to be fixed, which is the right instinct. The operator’s Engineering Framework work and its annual Electricity Statement of Opportunities both now explicitly model minimum demand troughs years ahead, and AEMO has been candid in its public statements that negative and near-zero demand periods are becoming a normal part of system operation rather than an edge case. You can read AEMO’s own forecasting methodology directly on its website at aemo.com.au, and the Clean Energy Regulator publishes the small-scale installation data that underpins most of this analysis at cer.gov.au.
The practical response has three legs: more synchronous condensers and grid-forming batteries to hold the grid stable when rooftop solar dominates; dynamic export limits to manage feeder-level voltage risk in near-real time rather than with blunt static caps; and tariff reform to actually price the two-way flow so household decisions line up with grid need. None of that is glamorous. None of it makes a good press release photo. But it’s the actual engineering answer, and it’s happening, just slower than the installation rate that caused the problem in the first place.
A quick word on negative prices, because they get misread #
Negative wholesale prices at midday get reported as a solar failure. They’re closer to the opposite: a price signal working exactly as designed, telling large generators to switch off because there’s too much supply for the demand on offer. I went through the mechanics in negative electricity prices at midday: what they mean for you. The problem isn’t that prices go negative. The problem is what happens to system security when almost everything synchronous has switched off in response, leaving the grid thin on inertia right at the moment cloud cover or a transmission fault could do real damage.
It’s a bit like watching a cricket side declare early because the pitch has flattened out completely: the economics of the moment make total sense, right up until someone needs a bowler who can actually turn the ball and finds the cupboard bare.
So is it breaking the grid? #
No, not in the sense of an imminent collapse. But it has broken the old assumptions the grid was planned around, and those assumptions were baked into network tariffs, inverter standards and connection agreements written when rooftop solar was a rounding error, not a quarter of the generation mix on a sunny Sunday. The fix isn’t fewer panels. It’s a grid, a market and a set of rules that finally treat distributed solar as the serious piece of generating capacity it’s become, with the obligations and the coordination that implies.
Whether the AEMC and state regulators move fast enough to get ahead of the next three gigawatts a year, rather than permanently chasing the last three, is the question I’ll keep asking.
– Marcus Wren, Editor
Photo by Evgeniy Alyoshin on Unsplash