Building Before Knowing
Snowy 2.0 may still matter to Australia’s energy future, yet its reported cost blowout shows the risk of turning early confidence into public commitment.
The hardest part of building something large is knowing enough before the work begins.
A note on context: This essay follows Andrew Probyn’s Sydney Morning Herald article, Snowy Hydro bracing for 10-fold cost blowout, published on 15 May 2026, and uses it as a starting point for a broader reflection on how large projects are promised, planned and judged.
Introduction
Snowy Hydro 2.0 was meant to be a confident answer to a difficult national question: how does Australia store enough renewable energy to keep the grid reliable when the sun is not shining and the wind is not blowing?
The answer offered in 2017 was bold and easy to understand. Connect two existing reservoirs through tunnels. Pump water uphill when electricity is abundant. Release it when power is needed.
It was nation-building with a clean energy purpose.
But the story now looks more complicated. The project was first presented with a $2 billion price tag and a 2021 completion date. It is now reported to be heading towards a much higher cost, possibly around $22 billion, with completion pushed towards late 2028.
This does not automatically mean Snowy 2.0 is a bad project. It may still prove useful, even necessary. But through the lens of Bent Flyvbjerg and Dan Gardner’s How Big Things Get Done, it raises a sharper question.
What happens when a project is publicly promised before it is properly known?
Snowy 2.0 may not be a failed project. It may be a failed promise.
The Promise Before the Project
Early certainty can become the most expensive part of a project.
Every big project begins with a story. Snowy 2.0’s story was especially powerful: the Snowy legacy, renewable storage, national scale, and the simple image of Australia’s biggest battery.
That story helped the project make sense. It also made the early numbers matter more than they should have.
The original public frame did three important things:
The first number became the anchor: The $2 billion figure shaped public expectation long after it stopped describing the project.
The first date created confidence: A 2021 completion date suggested the project was not only desirable, but ready.
The clean energy purpose may have softened scrutiny: Because the project aligned with the energy transition, early uncertainty may have been easier to overlook.
The article’s most important line comes from Snowy Hydro chief executive Dennis Barnes:
Although there was the famous $2 billion by 2021 announced, no work had been done at that point.
That sentence changes the whole story. It suggests the public promise came before the project had been properly understood.
The next issue is why that matters so much for large projects.
Thinking Fast, Acting Slow
The faster a commitment is made, the longer reality may take to answer back.
Flyvbjerg and Gardner argue that successful projects tend to “think slow, act fast”. They spend more time planning, testing and reducing uncertainty before full commitment. Once the plan is strong, execution can move quickly.
Troubled projects often do the reverse. They think fast and act slow.
Snowy 2.0 appears to carry that risk. The commitment was compelling, but the delivery has become slow, costly and exposed.
The pattern is familiar:
Planning appears to have lagged behind commitment: The early public confidence is hard to reconcile with later admissions about pre-feasibility assumptions and immature design.
Unresolved risks moved into construction: Geology, logistics, labour and access issues do not disappear when they are under-examined. They reappear later as cost and delay.
The window of vulnerability widened: A project meant to move quickly has instead remained exposed for years to technical setbacks, financing pressure and changing energy conditions.
This is not to say every problem could have been predicted. Large underground projects are genuinely difficult. But that is precisely why early certainty is dangerous.
That leads to another Flyvbjerg idea: the need to look beyond a project’s own story.
The Inside View and the Outside View
A project always looks simpler from inside its own story.
From the inside, Snowy 2.0 had a persuasive logic. It used existing reservoirs. It expanded a famous national asset. It promised renewable storage. It offered one large answer to a difficult grid problem.
But the outside view asks a colder question: what usually happens to projects like this?
That question would have highlighted three warnings:
This was a bespoke project: Remote tunnelling, underground caverns, difficult geology and national park logistics are not ordinary construction conditions.
The Snowy name may have created comfort: The original scheme’s legacy gave the project emotional authority, but it did not remove the risks of this new build.
Reference-class thinking was essential: The better question was not simply “can this be built?” but “how do similar projects usually perform on cost and schedule?”
Optimism is not always foolish. Public ambition often needs it. But optimism becomes risky when it is not disciplined by comparison.
And comparison matters most when a project is not modular.
Bespoke Risk and Lock-In
The harder something is to repeat, the harder it is to learn from cheaply.
Flyvbjerg and Gardner give weight to modularity. Projects usually perform better when they can be repeated, standardised and improved through iteration.
Snowy 2.0 is not that kind of project.
It is not a row of solar panels, a fleet of standardised batteries, or a repeatable rollout. It is a one-off underground civil engineering project shaped by geology, water, terrain, access and sequence.
That creates a particular kind of risk:
Learning is expensive: Once lessons are learned underground, after contracts and commitments are in place, every discovery can carry a high cost.
Progress does not remove all risk: Tunnel boring machine Florence’s 11-month delay shows how quickly underground conditions can disrupt planning assumptions.
Physical progress creates lock-in: Once tunnels are bored and caverns built, the question shifts from “should we build this?” to “how do we finish what we have started?”
This is the uncomfortable middle ground. Snowy 2.0 may be rational to complete even if the original promise was flawed.
A project can still be useful and still have been badly promised.
That brings us to the question that should sit at the centre of the current debate.
The Project That Changed While Keeping Its Name
A project can keep its name while becoming a different bargain.
Snowy 2.0 still has the same name, the same broad purpose and the same symbolic connection to national infrastructure.
But it is not the same public proposition.
A $2 billion project completed in 2021 is different from a much more expensive project completed years later. The strategic need for storage may remain. The case for Snowy 2.0 may still be strong. But the facts have changed.
That creates three tests:
The cost to complete must be understood: The immediate question is how much more money is needed to finish the asset.
The cost to justify must be tested: The deeper question is whether the project’s future benefits still justify its current cost, timing and risk.
The public case should be refreshed: If the bargain has changed, the justification should change too.
This is not an argument against big infrastructure. It is an argument for clearer judgement about big infrastructure.
The public was first asked to support one version of Snowy 2.0. It is now being asked to carry another. That difference deserves to be named.
Conclusion
Snowy 2.0 may still become an important part of Australia’s energy system. It may yet provide long-duration storage, strengthen reliability and help support a grid increasingly shaped by wind and solar.
But through Flyvbjerg and Gardner’s lens, it also shows the cost of early certainty. It shows how a confident public promise can become a trap when the project beneath it is still immature. It shows how the first number can anchor public consent long after it has stopped describing reality.
The lesson is not that Australia should avoid building big things.
It is that Australia should be more careful about how big things are promised.
Large infrastructure requires ambition, but it also requires humility. It requires leaders to say what is known, what is uncertain, what might change, and when a project’s public case needs to be tested again.
Snowy 2.0 is meant to store energy. For now, it is also storing a lesson about planning, confidence and the price of discovering reality too late.
The future is not built by certainty alone, but by the discipline to question certainty before it hardens into commitment.