AEMO's 2026 plan is an honest, lowest-cost design for the grid. But read it closely and it quietly assumes a future it never states out loud: the households who can afford to leave the grid will, and the households who can't will be left paying for a network that costs more every year. This is a plain read of where the headline saving actually went, why the rules will never catch up in time, and who is left paying the bill by 2035.
No serious person is arguing against cleaning up the grid. The argument is about how fast, in what order, and who pays if we get the order wrong. AEMO's plan is built to make the total cost as low as possible. It is not built to decide who ends up paying, or when. Those are political choices the plan treats as someone else's job. They are the whole problem.
One number in the 2026 plan went down. Almost everything underneath it went up: the cost to build the powerlines, the cost to run a household on them, and the reward for walking away. Hold those four lines in your head and the rest of this page is just the detail behind them.
Each line set to 100 in the 2024 plan · directional · the real figures are sourced in the sections below
A well-planned clean-up and a rushed one reach the same grid by 2050. They do not produce the same bill — and the bill does not fall to the same people.
Between the 2024 plan and the 2026 plan the headline fell from $122B to $106B Fact — a 13% saving, announced as progress. Here is what actually changed. AEMO lowered the discount rate it uses and counted some finished works, which makes future spending look smaller in today's dollars. That is real, but it is a sum on a spreadsheet, not a cheaper powerline. The concrete did not get cheaper. The powerlines got dearer. The real cost of building overhead transmission is running 25 to 55% higher per kilometre Fact. In the same 18 months, a roughly $5B interconnector was quietly downgraded from committed to maybe. Fact
Set the 13% saving next to the projects already being built and it looks less like discipline and more like a rounding error sitting on top of some very large overruns.
First or modelled estimate → latest estimate · you repay these through network charges for the life of the asset Fact
A headline that falls 13% because of a discount rate, sitting on projects that are 90 to 500% over budget — that is not a cheaper system. It is a more expensive system, described with a cheaper sentence.
When a transmission project goes over budget, the extra cost is not written off. It is added to the pile of assets the network company is allowed to earn a return on. The company then earns that return year after year — roughly 6 to 7% — plus the cost of slowly writing the asset off, plus maintenance set at 1.5 to 2% of the whole pile. Add it up and about 9 to 10 cents of every overrun dollar comes back on bills every single year, for the life of the asset. Inference Double a project's cost and you double its permanent maintenance charge too — about $98 to $170 a year per household on the big projects alone. Fact
That rising fixed charge then triggers two knock-on effects AEMO's lowest-cost model never prices in. First, as grid power gets dearer, anyone who can afford to put up their own solar and battery does — so grid sales fall, the same fixed costs are spread over fewer units, prices rise again, and more people leave. It is a vicious circle. Second, the rules that could break the circle move slowly. The pricing reform that might fix it does not arrive until around 2030 — roughly a decade after grid sales first started falling. Fact The cost keeps building faster than the rules can respond.
The two reinforcing effects are marked with a circular arrow · the rule that would stop them arrives late Inference
The official story blames solar households for pushing costs onto everyone else. The numbers say the real cost rocket is the transmission bill — and the "fix" of moving to fixed charges simply forces the households who can't leave to pay for powerlines the households who can are sensibly walking away from.
Here is the blind spot underneath all of it. Subsidies are fast and feel good, so governments hand them out early — cheaper home batteries, rooftop solar credits, the electric-car tax break. Roughly $3.85 billion a year flows this way, almost all of it claimable only if you own your home or salary-package a car. Fact The rules that would shape the system fairly — network pricing, who pays the fixed costs, protections for low-income homes — are slow. They run two to five years behind by the simple nature of how rule-making works: reviews, consultation, then a determination locked in for years. Fact
That gap is not a small timing nuisance. It is the whole risk. While the rules are still being written, the subsidies are already creating facts on the ground. Homeowners install, sink their capital, and become the winners. Renters and lower-income households become the losers — they helped pay for the subsidy pool through tax, but can't reach it. By the time the considered rules finally arrive, the winners are already in place, already invested, and now a voting bloc. The best available rule would have to claw value back from them — and a policy that has to take back from the people it just rewarded is far harder, slower, and angrier to pass than one written before the winners existed. Inference
Subsidies land years before the rulebook · by the time the rules arrive, the winners are set Inference
You can see it already in the detail. The electric-car tax break was meant to cost $90 million a year; it now costs $1.4 billion, heading for $3 billion — and because it works through the tax system, the higher your income, the bigger the saving on the very same car (about $25,000 for a high earner against $12,000 for a middle one). Fact Every month it runs unchanged, more people sign three-year leases and become a group that will resist losing it. The review isn't even due until mid-2027 Fact — by which point the winners are locked in for years.
Moving fast on subsidies before the rules are set isn't just risky — it quietly destroys the fair option. It creates a group of winners that any fair fix would then have to unwind. The cheapest, calmest moment to design this fairly is before the money goes out the door. We are spending that moment.
Around 2030 four timers go off together. About 65% of the NEM's coal is retired; transmission is running 18 to 24 months behind the plan it was meant to follow; the move to fixed network charges arrives without protection for low-income homes; and the clean-energy rebates that softened the early bills taper away. None of this is a guess — each one is already scheduled or already happening. Fact
The result is the first big affordability event of the clean-up: a credible 500,000-plus households in energy debt, quarterly bills crossing the $800 pain point, and parties campaigning against the transition polling 18 to 22%. Scenario 2030 isn't the disaster. It's the warning the system gets before the disaster — and the last cheap chance to change course.
of NEM coal gone, before firming and powerlines are fully in place. Fact
with quarterly bills crossing the ~$800 mark. Scenario
the politics of the bill starts to bite. Correlation
For most of the grid's history, fully cutting the cord was for off-grid farmers and true believers — paybacks of 24 to 33 years made it a poor decision. Fact Two trends cancel that out. Battery prices fall toward about $280 per kilowatt-hour by 2035, and electric cars double as a home battery on wheels, so a normal household setup — a 45 kWh home battery plus about 40 kWh of car backup — comfortably gets through three cloudy winter days in every NEM capital. Inference At the same time, the fixed charge climbs toward $910 to $1,142 a year. The maths flips.
Below about 3 years, leaving stops being niche and becomes a normal household decision Inference
The cruel symmetry: the households most able to leave are the ones who already own a roof, solar and an electric car — the better-off. Every one of them who leaves doesn't just stop paying; they hand their share of the fixed network bill to whoever is left. The grid doesn't collapse in 2034. It just quietly starts charging a smaller, poorer group for the same set of powerlines.
This is where the argument stops being about engineering and starts being about the kind of country the plan leaves behind. By 2035 the NEM splits into two energy economies. On one side, households with solar, a battery and an electric car, plugged into a shared trading scheme, paying an effective $400 to $900 a year. On the other, the grid-only households — roughly 30 to 35%, shut out of self-supply — paying $3,200 to $4,500 a year for the same electricity. Scenario A gap of four to eleven times, decided mostly by one thing: whether you owned a roof.
Same grid, two economies · the dividing line is who owns the assets, not who uses more power Scenario
Three things make this hard to undo once it sets. First, the reverse transfer: households without solar — about 6.1 million of them — help fund clean-energy subsidies through tax and network charges they can't claim, a net loss estimated at 15 to 30% of their spare energy spending. Inference · directional Second, the renters: 2.9 million households, a quarter to a third of all homes, at about 4% solar — shut out of both rooftop savings and the subsidies that only homeowners can claim. Fact Third, the politics: "fuel poverty" becomes a charged phrase and anti-transition parties settle at 15 to 22%. Correlation Past this point you can't fix it without openly moving money around — and the people who would fund that move have already left the grid.
The plan never says "we are building a two-country grid." It doesn't have to. It minimises total cost and lets the bill settle on whoever already lacks solar. The split isn't a fault in the plan. It's what the plan leaves out.
This isn't doom for its own sake. Run the next decade four ways and the finding points in one direction: the choice isn't between risk and safety — it's between risks managed, put off, or piled up. And the thing that decides which one you get is the same in every case. It isn't money or technology. It's whether AEMO, the rule-makers, the regulator and the network companies can actually move in step. Inference
| The four futures | How likely | Blackout risk 2027–32 | Spare firm power 2035 | What it is |
|---|---|---|---|---|
| Coordinated | 25–30% | 3.8% | +18.5 GW | Built on time; rules land. The good case — and not the likely one. |
| Delivered late | 40–50% | 4.8% | +0.1 GW | Right calls, late delivery. The most likely future. |
| Uncoordinated | 12–18% | 8.7% | −4.5 GW | Rule-making fails; the rooftop fleet runs unmanaged. |
| Coal held on, then dropped | 10–15% | 18.1% | −8.6 GW | Put off now; a harder fall after 2032. |
The likelihoods are the source analysis's own directional estimates, not AEMO outputs. Scenario
Read the second row again. The single most likely outcome — at 40 to 50% — is that we make the right calls and deliver them late. Late delivery is exactly the condition that bakes overruns into bills, drags the rules past 2030, and hands the gap to households. The most likely future is the argument on this page.
This is the part the word "adversarial" gets wrong. The point isn't to oppose renewables or net zero — it's to reach the same grid in 2050 for less money and fewer casualties. Four moves, none of them radical, most of them just done in a better order:
Pace transmission to what the workforce and supply chain can actually deliver, instead of compressing it into overruns you then repay forever. Same finish line, smaller bill. Inference
Swap homeowner-only rebates for a credit that follows the household and the clean power it uses — one a renter can claim too. This breaks the reverse transfer at its source, for about the same money. Inference
Write the pricing and fairness rules first, so support doesn't lock in a winners' group the fair fix later has to unwind. Order is the whole game. Scenario
Today's community-solar money reaches under 1% of renters. Covering about 500,000 households needs roughly $2 billion — small next to the overruns. Fact
Clean up the grid — yes. But build it in step so the bill is smaller, and write the rules first so it doesn't fall on the 2.9 million households that never had a roof to put solar on.
This page argues a position. It does not hide the seams. The project overruns, the subsidy figures, the coal-retirement timing, the frozen rules, the renter numbers and the four-futures model are directly evidenced Fact. The payback maths, the 9-to-10-cents recovery, and the lock-in argument are reasoned from that evidence Inference. The 2030 bill shock and the 2035 split are weighted views of the future, not predictions Scenario. The vote-share numbers move together with bills, but the cause isn't proven Correlation. The reverse transfer is a direction, not a precise figure — 15 to 30% is a range, pending household-level data.
Moving early makes the technology cheaper for everyone — the panels and batteries the whole world buys come down in price partly because countries like Australia bought them first, and that global benefit may dwarf the local unfairness. Going slower risks stranding investment, losing the skilled crews between waves of work, and leaning on gas to fill the gap. And unfairness can, in principle, be fixed with payments later — which is a policy failure, not a reason to slow the build. If you think the 2035 split is unacceptable, the cleanest answer might be to clean up faster and share the cost harder, not to slow down.
This page accepts most of that — and still comes down on the side of order and pace. Because the one thing the case-against assumes, and 40 to 50% of the four-futures model denies, is that we'll deliver on time and fix the unfairness later. We probably won't do either. So plan for the late case, and write the rules before they're impossible to write.
AEMO's number went down because the plan got cheaper to pay for over time — not because the grid got cheaper to build, or to live on. Follow that gap and it runs all the way to 2035: a grid that works, a plan that adds up, and a country quietly sorted into the people who own the sunlight and the people who pay the bill. That sorting isn't in the plan. It's in the silence around it — and in the simple fact that we hand out the money years before we write the rules.
Sources — AEMO 2026 Integrated System Plan and media release (25 Jun 2026); AER Transparency Review (7 Jan 2026); AEMC Pricing Review Final Report (Jun 2026); ALT·AEMO analysis bundle (system dynamics, grid-defection economics, the reverse-transfer study, and the NEM rooftop-energy scenarios). Likelihood and vote-share figures are directional estimates from the source analysis, not AEMO outputs. The SWIS (Western Australia) is left out throughout — it is not part of the NEM.
Walter is human and can make mistakes. This is independent analysis and commentary, not financial, legal or investment advice. No advice given — all care, no responsibility.