Snowy 2.0 has now cost past $12 billion and is still not generating a single watt. Meanwhile, a 200 MW / 400 MWh battery near Liddell in the Hunter Valley was energised in stages and dispatching into the NEM within months of final investment decision. Those two sentences contain almost everything you need to know about the pumped hydro versus big batteries debate — but they don’t settle it, because the two technologies are not really competing for the same job.
Let me be precise about that, because the power-versus-energy confusion does a lot of damage in this conversation.
Power and energy are not the same thing — and it matters here #
When someone says a battery is “200 megawatts,” they’re describing its power rating: how fast it can push electricity into the grid at any instant. When they say “400 megawatt-hours,” they’re describing its energy capacity: how long it can sustain that output before it’s empty. A 200 MW / 400 MWh battery runs at full power for two hours. That’s it.
Snowy 2.0, when it finally comes online, is rated at 2,000 MW with a storage reservoir that gives it roughly 350,000 MWh of usable energy. That’s not a rounding error — it’s a difference of three orders of magnitude in duration. At full dispatch it could run for more than 170 hours. No lithium battery project anywhere near that scale is on the horizon in Australia.
So the honest framing is this: batteries are brilliant at short-duration, high-cycling tasks. Pumped hydro is built for long-duration, seasonal shifting. The NEM needs both, and pretending otherwise is how you end up with bad policy.
What batteries actually do on the grid today #
Australia now has well over 3,000 MW of large-scale battery storage operating or under construction across the NEM, and the fleet is doing work that would have seemed implausible five years ago. The Hornsdale Power Reserve in South Australia — the original Tesla big battery, commissioned back in late 2017 — demonstrated early on that batteries could respond to frequency deviations faster than any gas peaker. That’s the sweet spot: frequency control ancillary services (FCAS), rapid arbitrage across the morning and evening ramps, and backstopping variable renewable output minute-to-minute.
AEMO’s 2024 Electricity Statement of Opportunities (ESOO) — and I’ll flag that the position has moved on since that document was published, with the 2025 ESOO now the reference document — identified short-duration storage as one of the most cost-effective tools for managing the steepening duck curve as rooftop solar continues its remarkable growth. If you want the numbers on that rooftop penetration story, our piece on Australia’s rooftop solar success has the detail.
The economics are genuinely good for batteries in the two-to-four-hour window. Capital costs for utility lithium iron phosphate systems have fallen hard over the past few years. Build times are short — sometimes under 18 months from financial close to first dispatch. And because batteries cycle daily, they accumulate revenue from FCAS and energy arbitrage at a rate a pumped hydro plant sitting half-empty for six months cannot match.
What pumped hydro actually does — and why duration is everything #
Pumped hydro works by moving water uphill when electricity is cheap (usually midday, when solar is flooding the market) and releasing it through turbines when electricity is scarce and prices spike. The physics are slow and the civil engineering is enormous, but the energy density of a full reservoir is unmatched by any electrochemical technology we have today.
The argument for pumped hydro is not about the next sunny afternoon. It’s about the six-day wind drought in winter, or the extended hot spell that drives demand for three weeks straight while output from solar drops overnight every night. Let’s be careful with that number, but AEMO’s Integrated System Plan 2024 (ISP 2024) identified a need for between 7,000 and 19,000 MW of firming capacity by 2050 depending on the scenario, with long-duration storage a core pillar of the preferred development path. Batteries alone cannot fill that role at any cost that makes sense — the sheer volume of cells required would be extraordinary.
The problem, as Snowy 2.0 has made painfully clear, is that pumped hydro is slow and expensive to build. The original budget for Snowy 2.0 was under $2 billion. The project has blown well past $12 billion by current estimates, with commercial operations now expected in the late 2020s. That’s not a reason to abandon the technology — it’s a reason to be more careful about site selection, contracting, and risk allocation on the next one. And yes, there will be a next one: both the NSW and Queensland governments have pumped hydro feasibility work underway at various sites.
A rough comparison — same capacity, very different jobs #
A worked example helps here. Imagine two projects, each rated at 250 MW of power output.
| Metric | 250 MW / 500 MWh battery | 250 MW pumped hydro (40-hour reservoir) |
|---|---|---|
| Storage duration | 2 hours | 40 hours |
| Energy capacity | 500 MWh | 10,000 MWh |
| Typical build time | 12–18 months | 7–15 years |
| Response time | Milliseconds | Minutes (start-up) |
| Best revenue stream | FCAS, daily arbitrage | Seasonal shifting, capacity payments |
| Geographic constraint | Low (grid connection only) | High (requires suitable hydrology and terrain) |
These are illustrative, not project-specific. But the ratios are real. You’d need twenty 500 MWh batteries to match the energy storage of that 40-hour pumped hydro plant — and you’d still get less than half the duration of Snowy 2.0’s full reservoir.
The market question nobody fully answers #
Here’s where I’ll offer a view that runs against the grain of a lot of industry commentary: I think the revenue model for long-duration storage in Australia is still genuinely unsolved, and building more pumped hydro before we fix that is a bet on government backstops rather than markets.
Batteries make money today because they cycle frequently and the FCAS and arbitrage revenues are real. A pumped hydro plant that sits mostly full for six months waiting for a winter cold snap needs a capacity payment or a contract that values its option value — its ability to dispatch when nothing else can. The NEM’s energy-only market design doesn’t naturally produce that signal. AEMO and the AEMC have been working on capacity mechanism designs for years; the position on that continues to evolve. Until there’s a durable revenue stream for very-long-duration storage, every pumped hydro project will require either government equity or a contracted off-take, and that shapes who can build them.
That’s not an argument against pumped hydro. It’s an argument for getting the market design right before we commit to the next $10 billion hole in a mountain.
For a broader sense of how firming fits into the whole renewable buildout, our explainer on the role of fossil fuels and renewables in Australia gives useful context. And if you want to understand where wind sits in this puzzle — because wind’s intermittency is half the reason we’re having this conversation — the piece on the rise of wind energy in Australia is worth a read.
Offshore wind adds another wrinkle #
One thing that changes the calculus slightly is offshore wind, which tends to generate more consistently through the night and in winter than onshore solar. If large offshore wind capacity comes online in Victoria and NSW over the 2030s — still conditional on licence awards and financing — the duration requirement for storage shifts somewhat. You still need long-duration backup, but the frequency of very-long droughts in net generation becomes lower. Whether that makes pumped hydro slightly less urgent or just differently sized is a question the ISP modelling is actively working through.
Where the two technologies actually fit together #
The honest answer to “pumped hydro or batteries?” is that it’s not either/or, and the NEM almost certainly ends up with a lot of both. Short-duration batteries handle the daily cycling, soak up the midday solar glut, and keep frequency stable. Long-duration pumped hydro — if it gets built on time and on budget, which is a genuine if — handles the multi-day events that would otherwise require gas peakers or demand curtailment.
ARENA and the CEFC have both backed battery projects at various stages; the federal government’s Capacity Investment Scheme has been the mechanism underwriting a lot of the pipeline. You can find ARENA’s project register and AEMO’s ISP documentation at their respective sites — both are worth bookmarking if you’re tracking the pipeline seriously.
I’ve spent a fair bit of time this year going through the ISP 2024 workbooks — something I do on long weekend rides when the numbers are stuck in my head and won’t leave — and what strikes me is how sensitive the modelled outcomes are to storage duration assumptions. Shift the assumed battery duration from two hours to four, and the role of pumped hydro shrinks noticeably. Shift it back, and pumped hydro becomes close to indispensable.
That sensitivity is the whole story, really. We are building a grid whose long-term shape depends on assumptions we’re still testing. The technology that firms it most effectively will be whichever one we manage to build, finance, and dispatch reliably — and on that score, the batteries are currently winning on points.
— Anjali Rao, Grid & Storage Correspondent
Photo by Jonathan Hanna on Unsplash