Drive past Lake Borumba on the Mary Valley road out of Gympie and, on the surface, not much has changed in two years. The lake still sits placid behind its existing wall. It’s the upper site, cut into the range above it, where the real work is happening: tunnel boring, access roads, the early civil works for what Queensland Hydro insists will be a 2,000 megawatt, 24-hour pumped storage scheme. Let’s be careful with that number, because “2GW” describes the power rating, not the energy the scheme can hold. Borumba’s design target is roughly 48,000 megawatt-hours of storage, which is the figure that actually tells you how long it can run flat out before the upper reservoir empties.
That distinction matters more here than almost anywhere else in the state’s pipeline, because Borumba is being sold, correctly, as a duration asset – something to cover the multi-day wind droughts and extended cloud cover that four-hour batteries simply can’t touch. I’ve written before about why that duration gap is the thing firming debates keep skating past, and Borumba is Queensland’s single biggest swing at closing it.
Where the money stands
The cost trajectory on this project has not been kind to anyone doing early-stage estimates. When the former Palaszczuk government first outlined Borumba in its 2022 energy plan, the working figure floated around $14.2 billion. By the time Queensland Hydro’s own project description and subsequent state budget papers caught up with deeper geotechnical work, that number had crept toward the high teens, and some public estimates now sit closer to $18 billion before a final investment decision has even been locked in. For a scheme that size, a few billion dollars of drift isn’t scandal, it’s the normal cost of doing serious geotechnical drilling in volcanic rock you’ve only mapped from the surface. But it’s also why I’d treat any headline capital figure for Borumba as a placeholder rather than a commitment, until Queensland Hydro or the state Treasury publishes a number tied to an actual contract award.
The state government’s own public statements have been reasonably upfront about this volatility, which I’ll give them credit for – plenty of proponents elsewhere in the NEM have been far less willing to update a number once it’s out in a press release.
The timeline question
Original scheduling had first power from Borumba around 2029-30. That slipped once under the Palaszczuk government’s own revised planning, and the incoming Crisafulli government’s energy review, handed down not long after the October 2024 state election, put a different lens over the whole Queensland pumped hydro programme, including the much larger (and now effectively shelved) Pioneer-Burdekin scheme. Borumba survived that review in a way Pioneer-Burdekin didn’t, but the public signal since has been a later completion window, generally discussed in the early-to-mid 2030s rather than the end of this decade.
That’s a meaningful shift for anyone trying to model Queensland’s capacity adequacy. AEMO’s Integrated System Plan leans on firm, dispatchable, long-duration storage arriving broadly on this kind of schedule to manage the retirement of the state’s coal fleet. Push Borumba out a few years and the ISP’s modelling assumptions start wearing thinner, not broken, but thinner, and AEMO has been explicit in successive ISP updates that project slippage is one of the biggest risks to the whole transition timetable, not just a Queensland problem.
Why the site is genuinely hard
Pumped hydro sounds simple on a diagram: two reservoirs, a tunnel, reversible pump-turbines, and it is simple in concept. The execution is where it gets brutal. Borumba needs a new upper reservoir built largely off-river, connected by tunnels bored through terrain that Queensland Hydro’s own geotechnical reporting has described as more variable than initial surveys suggested. Tunnel boring through inconsistent rock is slow, and slow is expensive, because you’re paying for a tunnel boring machine and its crew by the week regardless of how many metres you’ve actually advanced that week.
There’s also the access question. Borumba isn’t remote in the way the Snowy 2.0 site is remote, but it still requires substantial road upgrades and worker accommodation for a project of this scale, and some of that civil works spend doesn’t produce a single megawatt of capacity. It just makes the rest of the build possible. Anyone comparing Borumba’s headline cost per megawatt against a gas peaker’s build cost is comparing two very different categories of engineering, and I’d push back gently on commentary that does that comparison lazily, because duration-weighted storage and fast-ramp gas are solving different problems for the grid, not competing head-to-head for the same dollar.
Public ownership and the political backdrop
Borumba sits inside the broader architecture of Queensland’s energy plan, which I’ve covered in more detail elsewhere: the state’s preference for public ownership of large firming assets rather than leaving that risk to the market. That preference survived the change of government, which is worth noting given how much else in Queensland’s energy policy did get rewritten after the 2024 election. Queensland Hydro remains a state-owned entity, and the Crisafulli government’s public commentary has kept Borumba as the flagship long-duration project in its energy mix, even as it dialled back some of the more ambitious elements of the prior plan.
Where this gets politically interesting is cost recovery. A project in the high teens of billions, publicly funded, eventually needs a revenue model: regulated return, contracted capacity payments, or some hybrid, and that detail hasn’t been fully settled in public documents I’ve seen. Compare that to the Capacity Investment Scheme’s approach to underwriting private storage and generation investment elsewhere in the NEM, and you can see two quite different philosophies sitting side by side within the same national market.
What it means for the grid
Here’s the case for why Borumba matters beyond Queensland’s own borders. The state’s coal fleet is ageing, and Queensland’s own schedule for coal retirement assumes a credible replacement for firm capacity arrives on a broadly similar clock. Rooftop solar penetration in Queensland is among the highest in the country, which means the minimum demand problem (too much solar at midday, not enough firm supply at night or through a run of cloudy days) is sharper there than almost anywhere else in the NEM. A scheme that can store nearly two days of output at full discharge rate is one of the few technologies that genuinely addresses both ends of that problem: soaking up midday oversupply, then running for an extended stretch when wind and solar both underperform.
Batteries are doing excellent work on short-duration smoothing, and I don’t want to undersell what big battery projects have achieved on frequency response and arbitrage. But the four-hour duration that dominates grid-scale battery deployment right now simply isn’t built for a five-day wind lull. That’s the gap pumped hydro is meant to fill, and it’s the central argument in the broader pumped hydro versus batteries debate that keeps resurfacing across the NEM.
A worked example, because the numbers deserve it
Take Borumba’s design figures at face value for a moment: 2,000MW of power capacity and roughly 48,000MWh of energy storage. Divide the two and you get 24 hours of continuous full-power discharge, which is where the “24-hour” description comes from. In practice, no operator runs a scheme flat out for a full day very often. It’s more useful to think of Borumba cycling at partial output over multiple days, covering a dip in wind generation across a broader weather system rather than a single evening peak. That’s a genuinely different service to what a four-hour battery fleet provides, and it’s why direct cost-per-megawatt comparisons between the two technologies usually mislead more than they inform.
I’ll admit I used to find this kind of project easier to wave through on faith that megaprojects always eventually land somewhere close to plan. Years of watching Snowy 2.0’s own cost and schedule drift (a project I’ve tracked closely, past its own multi-billion-dollar cost increase) cured me of that. Borumba isn’t Snowy 2.0’s twin, the geology and the ownership structure are different, but the pattern of early estimates proving optimistic is depressingly consistent across pumped hydro globally, not just in Australia.
The read from here
My own judgement, for what it’s worth: Borumba will get built. The site work is too advanced and the political commitment too entrenched, across two different state governments now, for it to go the way of Pioneer-Burdekin. What I’m less confident about is the state hitting anything like its original completion window, or holding the cost line anywhere near the numbers first floated in 2022. Readers modelling Queensland’s capacity adequacy off AEMO’s published ISP assumptions should build in some slack on both fronts, because the project’s own history over the past three years gives you the evidence to do exactly that.
There’s a decent chance Borumba ends up being judged less on whether it arrives on the original schedule, and more on whether, once it’s finally spinning, it actually does the two-day firming job the whole plan was built around. That’s a question for the early 2030s. For now, it’s tunnel boring machines and geotechnical reports, and a lake near Gympie that looks exactly the same from the Mary Valley road as it did two years ago.
– Anjali Rao, Grid & Storage Correspondent
Photo by American Public Power Association on Unsplash