Past $12 billion. That’s the number Snowy Hydro’s owner, the Commonwealth, has to make peace with before anyone can answer the question this piece is asking. Will Snowy 2.0 ever earn its keep? Not “is it a good idea in theory” – that ship sailed around 2017, but will the cash actually come back, on a timeframe that matters to the people who paid for it. I’ve spent a fair bit of this year going back through Snowy Hydro’s own corporate updates and the Australian National Audit Office material on the project, and the honest starting point is that the cost curve and the revenue case have been moving in opposite directions for years now.
Start with what the project actually is, because the public debate still gets this muddled. Snowy 2.0 is a pumped hydro scheme linking the Tantangara and Talbingo reservoirs via roughly 27 kilometres of tunnel, built to shift 2,000 megawatts of generating capacity and about 175 hours of storage into the grid. It is not a power station in the conventional sense. It is a giant battery that buys cheap electricity, pumps water uphill, and sells it back expensive. The economics of every pumped hydro asset in the world rest on one thing: the spread between the price you pay to fill the dam and the price you get for draining it. Get that spread right and the asset prints money for decades. Get it wrong and you’ve built the most expensive hole in the ground in the Southern Hemisphere.
The cost trajectory nobody wanted to admit
Snowy 2.0 was originally pitched under the Turnbull government at roughly $2 billion. That figure was always soft. It came before a geotechnical survey had properly characterised the rock the tunnel boring machines would be chewing through. By the time Snowy Hydro locked in major construction contracts the estimate had moved past $5 billion, and it kept moving. The project has now blown through $12 billion in capital cost, and that figure excludes the transmission build needed to actually get the power to market – HumeLink and the rest of the connecting infrastructure that EnergyCo and Transgrid are still working through.
The tunnel boring machine known as Florence became something of a national punchline along the way, stuck for extended periods in squeezing ground conditions under the Snowy Mountains that the original geotechnical work didn’t fully anticipate. Delays compound. Every extra year of construction is another year of finance costs accruing on a project that isn’t generating a cent of revenue. That’s not a scandal, it’s just how big tunnelling projects behave when the rock doesn’t cooperate, but it does mean the numbers used to justify the project in 2017 bear almost no resemblance to the numbers an investor would use today.
The revenue case was built for a different grid
Here’s where I think the public conversation goes wrong. Most of the criticism of Snowy 2.0 focuses on the blowout – fair enough, it’s a big number. Less attention goes to whether the grid Snowy 2.0 was designed for in 2017 is the grid it will actually operate in once it’s finally commissioned.
In 2017 the case for a 175-hour pumped hydro asset was built around daily and weekly arbitrage – storing cheap overnight coal power and selling it into evening peaks, with some seasonal shifting thrown in. That’s a sound role for long-duration storage. But the grid has moved a long way since then. Rooftop solar now routinely pushes wholesale prices negative across the middle of the day in most mainland states, a dynamic I’ve written about before when looking at the midday minimum demand problem. Big batteries (the four-hour, two-hour, even shorter duration assets that Akaysha, Neoen and others have built across the NEM) have already captured a huge share of the daily arbitrage trade that Snowy 2.0 was originally designed to dominate.
That doesn’t make Snowy 2.0 worthless. Four-hour batteries can’t do what a 175-hour asset does during an extended wind drought or a run of overcast winter weeks. But it does mean the easy money, the daily charge-discharge cycle that pumped hydro has lived on for a century, is getting competed away by assets that cost a fraction as much and took a fraction as long to build. Projects like the Waratah Super Battery are already proving there’s real commercial value in sub-day firming without anyone needing to bore 27 kilometres through granite.
Who actually pays if the spread doesn’t show up
Snowy Hydro is wholly owned by the Commonwealth, which matters more than most coverage lets on. If Snowy 2.0 were a private merchant asset, a shortfall in revenue would show up as a write-down, a shareholder complaint, maybe a credit rating problem. Because it sits on the Commonwealth’s balance sheet, a shortfall instead shows up as an opportunity cost buried in budget papers, visible only to people who go looking for it. That’s a structural problem I think gets underweighted in the debate. Public ownership can build things private capital won’t touch: our piece on public versus private generation ownership goes into that trade-off, but it also means the bill for a disappointing return doesn’t land on a clearly identifiable party. It lands on taxpayers generally, diffusely, for a long time.
Follow the money and you find the real test isn’t whether Snowy 2.0 generates revenue. Almost any dispatchable asset generates some revenue in the NEM’s current price settings. The test is whether that revenue, discounted back over thirty or forty years, comes close to justifying a capital bill that’s now several multiples of the original estimate. On the numbers as they stand, I don’t think it does, not without some combination of capacity payments, contracted revenue through the Capacity Investment Scheme, or a materially different wholesale price environment than the one we’re in now.
The system-value argument, and why it’s not nothing
Snowy Hydro and AEMO’s own planning documents make a different case, and it deserves a fair hearing. The Integrated System Plan treats long-duration storage as a structural requirement for a grid that’s shedding coal faster than most states can build replacement firming: a mismatch I’ve covered before looking at how NSW’s coal exit is outrunning its REZ build. Under that framing, Snowy 2.0 isn’t meant to be judged purely as a merchant trading asset chasing the best daily spread. It’s meant to be judged as system insurance: the thing that keeps the lights on through a week-long wind lull in winter when every battery in the state has already cycled flat.
That’s a real function. Pumped hydro versus batteries isn’t a question with one right answer – different durations solve different problems, and I’ve gone through that trade-off in more detail elsewhere on this site. The honest problem for Snowy 2.0 is that insurance value is genuinely hard to monetise in a market built around five-minute dispatch pricing. AEMO can model the reliability benefit. Getting the market to actually pay for it, dollar for dollar, through a mechanism that doesn’t ultimately route back to consumers or taxpayers, is a different exercise entirely.
What commissioning actually looks like from here
Snowy Hydro’s own updates have pushed full commissioning out repeatedly, and at this point I’d treat any firm date with real scepticism until tunnelling is fully complete and the underground machine hall is proven. The project needs HumeLink energised to actually export power at scale, and that transmission build has its own approvals and landholder friction to work through: the kind of fight that’s become familiar anywhere a new line needs easements, as anyone following the VNI West dispute in Victoria’s north will recognise. Even once Snowy 2.0 is physically finished, it doesn’t instantly become commercially useful. It becomes commercially useful once the wires exist to move its output to where demand actually sits.
That stacking of dependencies (tunnel, machine hall, transmission, market design) is why I think the project’s defenders undersell the risk and its critics oversell the stupidity. Neither extreme is doing the arithmetic properly. The capital is sunk either way at this point; walking away now would crystallise the loss without getting any of the system benefit. The real decision in front of policymakers isn’t whether to finish Snowy 2.0. It’s what revenue architecture gets built around it once it’s running, because left purely to five-minute wholesale pricing, I don’t think the spread shows up often enough to justify the bill.
A quick aside, because it’s hard to resist
There’s a cricket analogy here I can’t quite let go of. Snowy 2.0 is the batter who’s spent four days compiling a painstaking double-century against bowling that’s since changed completely: the declaration never came, the follow-on’s been and gone, and the scoreboard looks impressive until you ask what match situation it was actually built for. The innings might still matter. It just won’t be judged on the terms it was played under.
Where this leaves the project
My own view, for what it’s worth: Snowy 2.0 will eventually generate real revenue, because a 2,000 megawatt, 175-hour asset connected to the NEM’s biggest demand centres isn’t going to sit idle. The question was never “will it earn anything.” It’s whether it earns enough, on a cost base that’s roughly six times the original pitch, to count as a good use of $12 billion-plus in public capital rather than merely a necessary one. I don’t think the current market design gets there without further intervention: some flavour of long-term contract or capacity payment that effectively nationalises the insurance value AEMO’s planning documents already ascribe to it. That’s not a crisis. It’s just a different conversation to the one the project was sold on in 2017, and I’d rather the public debate caught up with that than keep re-litigating the tunnelling delays as though they’re the whole story.
For readers wanting the fuller account of the project’s engineering and contract history, our earlier profile on Snowy Hydro as a government-owned giant juggling contested projects is worth the read alongside this one. The numbers keep moving. The question underneath them doesn’t.
Further detail on Snowy Hydro’s own project reporting is available through its corporate updates, and AEMO’s Integrated System Plan sets out the long-duration storage assumptions referenced above: both are public documents worth reading in full rather than taking on faith from either side of this debate.
– Marcus Wren, Editor
Photo by American Public Power Association on Unsplash