Book launchLaunched 10 October at CPAC, BrisbaneDetails
Sandra BourkeAuthor of To Save a Beach
Resources: national snapshot, as at August 2026

Australia’s Pumped Hydro

Australia runs three pumped hydro schemes, the newest finished in 1984; the one it is building has gone from $2 billion to $12 billion and is not finished, and the one Queensland has been promised would cost more than both and has no decision to build it.

The Scale

OperatingThree schemes, 2,610 megawatts: Tumut 3 (1973), Shoalhaven (1977), Wivenhoe (1984). Together they can push nine-tenths of Eraring’s power, but only for hours, and only after using more electricity to pump the water up than they get back.
Being builtSnowy 2.0: 2,200 megawatts, 350,000 megawatt-hours, 27 kilometres of tunnel, a power station 800 metres underground, about 5,000 workers. Full, it could run at full power for 159 hours, six and a half days. Its stored energy equals nine days of Eraring’s output.
Its costAnnounced in March 2017 at about $2 billion for completion in 2021. Contracted in 2019 at $5.1 billion. Reset in August 2023 to $12 billion and December 2028. A nine-month cost reassessment ordered in October 2025 had not reached the minister by August 2026.
Also being builtKidston, in two flooded pits of an old north Queensland goldmine: 250 megawatts, 2,000 megawatt-hours, eight hours at full power, $777 million, under construction since May 2021, now slated to generate in July 2027.
ProposedBorumba, in the Mary Valley: 2,000 megawatts, 48,000 megawatt-hours, 24 hours at full power, a day and a quarter of Eraring’s output. Costed at $14.2 billion in June 2023 and $18.4 billion in December 2024, first power moved from 2030 to July 2033, completion to July 2035. No decision to build.
Cancelled and shelvedPioneer-Burdekin, 5,000 megawatts and 120,000 megawatt-hours, cancelled in November 2024 at an announced $12 billion. Mount Rawdon, 2,000 megawatts and 20,000 megawatt-hours in a worked-out goldmine, $50 million allocated, no decision.
Still requiredAEMO’s plan needs 5 gigawatts of deep storage from new pumped hydro by 2050. It counts Snowy 2.0 and Kidston as 2.5 gigawatts by the end of 2028 and Borumba as 2 gigawatts “anticipated”.

What Exists Now

Tumut 3 in the Snowy has run since 1973 and remains the largest at 1,800 megawatts, with six generating units and three pumps. Shoalhaven, 240 megawatts at two sites in Kangaroo Valley, followed in 1977. Wivenhoe, 570 megawatts west of Brisbane, came in 1984 and holds about ten hours of water in its upper pond. Nothing has been finished since.

What Is Being Built

Snowy 2.0 links Tantangara and Talbingo, two reservoirs of the old scheme, through 27 kilometres of tunnel, about twelve Sydney Harbour Tunnels laid end to end, with six reversible turbines in a cavern 800 metres down. Four tunnel-boring machines have been at work; the first broke through into the cavern in May 2026.

In June 2026 the Auditor-General found Snowy Hydro only partly effective in managing delivery, with no agreed baseline schedule and no robust understanding of the forecast cost to complete; the project was 262 days behind at July 2025, with a line of sight to September 2029. Snowy Hydro’s public page still says complete at the end of 2028. Its own annual report targets progressive commissioning in 2030. The Commonwealth Budget carries Snowy 2.0 as a fiscal risk and a contingent liability with no dollar figure attached.

Snowy 2.0 at $12 billion works out at about $5,500 for each kilowatt of power and about $34 for each kilowatt-hour of storage, because it holds so much water. Borumba at $18.4 billion works out at $9,200 a kilowatt and $383 a kilowatt-hour. The CSIRO’s 2026 benchmark for a 24-hour scheme was reported as $5,687 a kilowatt, or $237 a kilowatt-hour.

What Remains Proposed

Borumba would raise a new dam wall 300 metres downstream of the existing Borumba Dam and build a new upper reservoir on the ridges about 330 metres above it, joined by tunnels to an underground powerhouse. Announced in 2021, it was reviewed after the October 2024 election; the review found less than a 1 percent chance of first power by the promised 2030. In January 2026 the Coordinator-General extended the project declaration to February 2028. The 2026-27 state budget gave Queensland Hydro $324.3 million for exploratory and early works, subject to a continuing review. The refreshed business case is due in 2026.

Twelve other schemes remain proposed, 8,485 megawatts between them. None has a decision to build. Two New South Wales proposals were withdrawn in 2025: Origin’s Shoalhaven expansion and the 325 megawatt Central West scheme.

How Long the Water Lasts

SchemePower and hours at full output before the upper reservoir is empty
Snowy 2.0 (building)2,200 megawatts for 159 hours
Borumba (proposed)2,000 megawatts for 24 hours
Pioneer-Burdekin (cancelled)5,000 megawatts for 24 hours
Mount Rawdon (proposed)2,000 megawatts for 10 hours
Wivenhoe (operating)570 megawatts for about 10 hours
Kidston (building)250 megawatts for 8 hours
Tumut 3 and Shoalhaven (operating)hours not published

How Long They Take

Tumut 3 took about five years to 1973. Wivenhoe took about seven, from 1977 to 1984. Kidston, begun in May 2021, will have taken more than six years if it generates in July 2027. Snowy 2.0, contracted in April 2019, will have taken nine to eleven years on the dates now in circulation. Borumba’s own review puts completion 14 years after its 2021 announcement.

Reading the Numbers

Pumped hydro generates nothing. It moves electricity from one hour to another and gives back less than it takes, so it adds nothing to a year’s supply and is never compared with Eraring’s output. Its megawatts say how hard it can push; its megawatt-hours say for how long. A feasibility study, a business case or an exploratory-works allocation is not a decision to build, and none of the proposed schemes here has one. Kidston is registered with AEMO but had neither pumped nor generated at the status date, so it is counted as under construction.

Technical material prepared for To Save a Beach and removed from the manuscript for length. Figures are as at August 2026. Where a figure is an estimate, the assumption is stated.