Are we closing the power stations before the replacements are built?

Australia still gets a large share of its power from coal. In 2025, coal made 43% of the electricity in the National Electricity Market, the grid that runs from Queensland to South Australia and Tasmania (AEMO figures reported by Enerdata, 27 August 2026). Over the next ten years about 15 gigawatts (GW) of coal and gas generation is scheduled to close. Over the same period, electricity use is forecast to rise by more than 40%, as homes and cars electrify and data centres grow (AEMO, 2026 Electricity Statement of Opportunities).

So the fear is simple. If the old stations shut before the new ones are ready, prices spike and the lights flicker. This page tests that fear against the market operator's latest numbers. It adds to our existing pages on the real cost of the transition, gas and energy security and the transmission build-out, rather than repeating them.

Which power stations are closing, and when?

The dates below are the ones AEMO used in its 2026 Electricity Statement of Opportunities (ESOO), published on 25 August 2026. They come from the owners' own notices to the market.

• Yallourn, Victoria (about 1.45 GW), mid-2028. EnergyAustralia announced this in March 2021, bringing closure forward from 2032. Yallourn burns brown coal in the Latrobe Valley.
• Gladstone, Queensland, March 2029.
• Eraring, NSW (about 2.9 GW), April 2029. Australia's largest power station. Origin first planned to close it in August 2025. The NSW Government struck a deal in 2024 to keep it open to 2027, and in January 2026 Origin extended it again to April 2029. Our Eraring page covers that deal in detail.
• Bayswater (2.7 GW) and Vales Point B (1.3 GW), NSW, 2033.
• Loy Yang A, Victoria (2.2 GW), 2035.

Yallourn, Eraring and Gladstone together produced an average of 29.3 terawatt-hours a year over the past four years. That is about 15% of all the electricity in the market, and all three are due to go within about nine months of each other (AEMO ESOO, as reported by The Energy, 25 August 2026).

One state is moving the other way. Queensland's LNP government announced in April 2025 that Callide B, a 700 megawatt plant built in 1988, would run past its planned 2028 closure to at least 2030 or 2031 (RenewEconomy, 8 April 2025).

Who decides when a coal station closes?

This point matters for fairness. In NSW and Victoria the coal stations are privately owned. Governments of both parties sold them over the past two decades. The closure dates are commercial decisions by Origin, EnergyAustralia, AGL and others. No federal law orders a coal plant to shut on a given day.

Policy still shapes those decisions. Emissions targets, subsidies for new renewables and cheap daytime solar all squeeze the income of old coal plants. But a government cannot simply order a private owner to keep running a plant that loses money. It can pay them to stay, as NSW did with Eraring. That is the main lever available, and it was used.

What does the market operator say about a reliability gap?

AEMO's 2026 ESOO is better news than the last few. It forecasts no reliability gaps anywhere in the market before 2030. AEMO credits a record 9.1 GW of new generation and storage connected in 2025-26, and a pipeline of about 40 GW of committed and anticipated projects due by the early 2030s (AEMO, 2026 ESOO).

That is the headline. The detail is less comfortable. AEMO runs two main cases. The first counts only projects that are committed or anticipated, meaning they have money, approvals or contracts in place. On that basis, NSW and Victoria first breach the reliability standard in 2030-31. South Australia follows in 2031-32, Queensland in 2032-33 and Tasmania in 2033-34 (The Energy, 25 August 2026).

The second case adds government-backed schemes and "actionable" transmission projects that are planned but not yet built. That pushes the NSW, Victorian and South Australian gaps out by two to three years. In other words, the safety margin after 2030 depends on projects that have not yet reached a final investment decision.

AEMO calls the difference a "deliverability gap". It is widest for wind, where 10.5 GW separates the two cases (Modo Energy analysis of the 2026 ESOO, 31 August 2026). NSW and Victoria account for 80% of the forecast 2036 shortfall in the government-backed case. AEMO's chief executive Daniel Westerman put it plainly: "Beyond 2030, the next wave of investment will be critical to maintaining reliability."

Put the dates side by side. Eraring is due to close in April 2029. On committed projects alone, NSW breaches the standard in 2030-31. That is a buffer of roughly 18 months, which is not much for an industry where big projects routinely slip by years.

The closures run on a fixed clock. The replacements run on a project schedule, and the project schedules keep slipping.

Why are the replacements running late?

Three projects show the pattern.

Snowy 2.0. This pumped hydro scheme is meant to be the biggest battery in the grid, about 2.2 GW and 350 gigawatt-hours of storage. It was announced in 2017 at about $2 billion (RenewEconomy, 7 November 2022). A reset in 2023 put the cost at about $12 billion. AEMO now assumes it starts in December 2028 (AEMO ESOO, as reported by The Energy). In June 2026 the Australian National Audit Office found that the final price and completion date were still unknown, and that "a reliable cost forecasting system is still not developed" (The Mandarin, 19 June 2026). Snowy 2.0 began under the Turnbull Coalition government. The current government inherited it and has kept funding it.

Offshore wind. Victoria set targets of 2 GW of offshore wind by 2032, 4 GW by 2035 and 9 GW by 2040. The first auction was meant to open in September 2025. It was postponed to August 2026, citing uncertainty over ports, transmission and market design (RenewEconomy). In 2025, three developers handed back their feasibility licences, including the AGL-led Gippsland Skies project (Daily Cargo News, 9 December 2025). Victoria's Auditor-General warned that no port had been approved to assemble the turbines, and that "under an optimistic scenario Victoria could build two gigawatts of offshore wind capacity by the end of 2033". The auction finally opened on 26 August 2026, with contracts due to be awarded in 2028 (Maritime Executive, 26 August 2026). On that timetable the 2032 target is very hard to meet.

Transmission. New wind and solar farms are useless to cities until high-voltage lines connect them. Our transmission page sets out the delays in detail. In Queensland, the CopperString line to Mount Isa grew to an estimated cost of nearly $14 billion, and the state cut it back to its first stage (RenewEconomy, 8 April 2025).

None of these are imaginary problems. But it is worth being fair about the scale. A record 9.1 GW was connected in a single year. Batteries are being built very quickly: the committed and anticipated pipeline includes 18.9 GW of grid batteries (AEMO ESOO, as reported by The Energy). The slow parts are the very large, one-off projects. The fast parts are batteries and solar.

Is the east coast running out of gas?

Gas matters to this story twice. Homes and factories burn it directly, and gas power stations are the backup that steps in when coal goes and the wind drops.

AEMO's 2026 Gas Statement of Opportunities (GSOO), released in March 2026, found the near-term picture had improved since 2025. Even so, it forecast anything from a 12 petajoule shortfall to a 3 petajoule surplus for the third quarter of 2026, depending on how much spare gas the Queensland LNG exporters sell at home. The southern states, Victoria, NSW, South Australia, Tasmania and the ACT, needed extra gas every month from April to September 2026 (Energy Magazine, reporting the GSOO).

The longer view is worse. AEMO warns of a risk of shortfalls on peak winter days in the southern states from 2028, and annual supply gaps from 2029. The main cause is the Longford gas plant in Victoria. It has historically supplied about two-thirds of the gas used in the east coast market, and its output is falling before it retires at the end of 2033. AEMO's Alice McLaren summed it up: "We're seeing an improvement in our forecast compared to last year… but longer term, those risks remain" (AEMO On Air, 30 March 2026).

Notice the timing. Peak-day gas risk arrives in 2028, the same year Yallourn closes. Annual gas gaps arrive in 2029, the same year Eraring closes. The backup fuel gets tight just as the backup is most needed. One new gas plant helps: AEMO assumes the 750 MW Kurri Kurri station in the Hunter is available from September 2026 (AEMO ESOO, as reported by The Energy). Plans to import gas through a floating terminal at Geelong have been delayed, with Viva Energy adjusting its investment timeline (Daily Cargo News, 9 December 2025).

Does Victoria's gas connection ban make the shortage worse?

Since 1 January 2024, new homes in Victoria that need a planning permit cannot connect to mains gas. The policy and its costs for households are covered on our Victorian gas ban page.

On supply, the ban works in the direction its supporters claim. It reduces gas demand. AEMO's 2026 Victorian Gas Planning Report Update forecasts Victorian demand on a typical winter day falling from 958 to 836 terajoules between 2026 and 2030. The catch is that Victorian production is falling faster still, so the shortfall grows anyway, to a forecast 40 petajoules a year by 2030 and 115 by 2035 without new supply (as reported on our gas ban page).

So the ban does not cause the shortage. Falling Bass Strait production does. But the ban also shifts more winter heating onto the electricity grid, in the same years that Yallourn closes. That means more electricity demand on cold, still winter evenings, which is exactly when solar is gone and wind may be weak. That trade-off is real, and the state's planning documents do not make it prominent.

What is happening to power prices?

This is where the popular story and the data part ways, so it needs care.

Wholesale prices fell sharply in 2026. In the June quarter of 2026, wholesale prices across the market fell by an average of 47% on a year earlier. Victoria fell 60% and NSW 53%. In NSW the average volume-weighted price dropped from $188 to $78 per megawatt-hour. Renewables supplied a record 42.1% of generation, and gas generation fell 30% (AEMO Quarterly Energy Dynamics, reported by the Newcastle Herald, 28 July 2026).

Regulated default offers fell from 1 July 2026. The Australian Energy Regulator's Default Market Offer fell between 3.4% and 5.0% in NSW and by 7.2% in south-east Queensland, but rose 1.4% in South Australia (Canstar, 26 May 2026). Victoria's Default Offer fell by an average of 5%, or $84 a year, to $1,591 for a typical household (Essential Services Commission, 25 May 2026).

But the bills people paid went up. The ABS found electricity costs were 22.4% higher in the year to June 2026, and 13.2% higher in the year to August. The ABS explains that government rebates, which had been taking money off household bills, have ended (ABS, 29 July 2026; EnergyPlans, 30 September 2026). Underlying prices eased, but the subsidy that hid earlier rises was withdrawn. Our page on the $275 promise tracks the gap between that pledge and household bills.

The cost risk from closures is mostly ahead, not behind. Wholesale prices spike when supply is tight. The 2022 crisis showed that. If Eraring and Yallourn go before enough firm capacity arrives, the cheap 2026 quarter will not be the guide to 2029.

What did governments get right?

Three things deserve credit. First, extending Eraring was a prudent hedge, made by a Labor state government against its own targets. Second, the record 9.1 GW of new capacity in 2025-26 is real and is the main reason the near-term outlook improved. Third, the regulated price benchmarks fell in 2026 in most regions, and wholesale prices with them. Critics who said renewables would push every price up every year have not been borne out by the 2026 data.

Would nuclear or keeping coal fill the gap?

This is context, not a recommendation. The Coalition took a plan to the May 2025 election for seven nuclear reactors across five states by the mid-2030s (AAP, May 2025). Whatever its merits, that timeline would not have produced power before Yallourn, Eraring and Gladstone close in 2028 and 2029. It speaks to the 2035 problem, not the 2029 one. Our nuclear ban page covers the legal and cost debate.

Keeping coal running longer is the other option, and Queensland is testing it with Callide B. That buys time, but ageing plants break down more often, and the Queensland Conservation Council, an environmental group, estimated that keeping Callide B open past 2028 could cost taxpayers up to $420 million (reported by RenewEconomy). That is an advocacy estimate and should be read as one.

The verdict: tested claim by claim

Claim: "We are closing coal plants before the replacements are built." Partly supported. Before 2030, AEMO says enough replacement capacity exists. After 2030, the replacements needed to stay within the reliability standard are not yet committed, and several flagship projects are years late.

Claim: "The lights will go out when Eraring closes." Not supported on current forecasts. AEMO sees no gap before 2030. The buffer after Eraring's April 2029 closure is thin, about 18 months on committed projects.

Claim: "The government is forcing coal plants to close." Not supported. Closure dates are set by private owners. Policy shapes their economics, but the one state that intervened, NSW, did so to keep a plant open.

Claim: "Snowy 2.0 is massively over budget and late." Supported. From about $2 billion in 2017 to about $12 billion in 2023, and the audit office says the final cost is still unknown.

Claim: "The east coast is running short of gas." Supported for the southern states in winter now, with AEMO warning of peak-day risk from 2028 and annual gaps from 2029.

Claim: "Renewables are making power more expensive." Not supported for wholesale prices in 2026, which fell sharply. Household bills rose because rebates ended, not because underlying prices did. The longer-term network costs of the build are a separate question, covered on our transition cost page.

For the wider picture, read renewables, firming and the real cost of the transition, gas, energy security and the interventions and the poles and wires the targets depend on.