Policy & Markets

Public versus private generation: who builds it better

23 November 2026 · by Marcus Wren
7 min read·1559 words·Updated 23 Nov 2026

$14.2 billion. That’s the latest publicly disclosed capital estimate for Queensland’s Borumba pumped hydro scheme, a state-owned asset being built by Queensland Hydro with no private equity partner anywhere near the capital stack. Compare that with Tilt Renewables, a wind and battery developer now owned by AGL and a consortium of pension funds, building out Rye Park and Oakey’s Clermont stage with a return hurdle set by its owners’ cost of capital, not by a state budget papers process. Two models, two very different ways of answering the same question: who should own the machines that keep the lights on.

The honest read is that Australia never actually chose between these models. We backed into both at once, state by state, and we’re now running the comparison live, in real time, with real money.

Why this debate resurfaced #

For twenty years the orthodoxy was simple: privatise the poles and wires, privatise the generators, let the market sort out what gets built. Victoria did it hardest in the 1990s. NSW followed more slowly, keeping networks public longer than generation. Queensland and WA never fully let go, and that decision now looks less like nostalgia and more like strategy.

Queensland’s energy plan leans on two pillars, pumped hydro and public ownership, and I’ve written before about how much weight that puts on one project succeeding. The state government effectively decided that firming capacity at the scale the grid needs by the early 2030s wasn’t something it could leave to a private developer’s investment committee. That’s a judgement about risk appetite as much as ideology. Private capital generally won’t fund an asset with an eight-to-ten-year build horizon and no merchant revenue certainty unless the return clears a steep hurdle. Governments can absorb that uncertainty because they don’t answer to a quarterly results call.

The case for private ownership #

Follow the money and the private model’s strongest argument is speed of capital deployment, not necessarily speed of construction. AGL, Origin, and increasingly the super-fund-backed developers like Tilt Renewables can move fast when a project clears their hurdle rate, because the decision sits with a handful of people rather than a cabinet process. Tilt’s build-out at Dundonnell, Rye Park and its Latrobe Valley battery assets happened without a single state election cycle derailing it. AGL versus Origin as gentailer strategies shows two different private approaches to the same transition, both moving largely on their own balance sheets, both answerable to the ASX rather than a parliament.

Private ownership also tends to price risk more honestly. A private generator that gets its capacity factor wrong, or misjudges firming costs against gas peakers, wears the loss on its own books. There’s a discipline in that. Capacity factor is the single most misunderstood number in this entire debate, and private boards tend to interrogate it harder than public ones because their lenders force the question.

The weakness is well known too. Private capital won’t build what doesn’t pay, and some assets the grid needs (long-duration storage, certain transmission links, remote network augmentation) don’t pay on any reasonable private timeframe. Marinus Link and HumeLink are both proof of that; both are public or public-adjacent precisely because no private sponsor would carry the completion risk at the published cost blowouts.

The case for public ownership #

Queensland’s bet on Borumba and the broader Queensland Hydro programme is the cleanest live test of public ownership at scale in the NEM right now. The state owns the risk, the state owns the construction overruns, and crucially the state owns the asset once it’s built, meaning the revenue doesn’t need to clear a private discount rate to be worth building. That’s the whole argument in one sentence: public ownership can justify infrastructure that earns its keep over forty years, not eight.

Victoria’s SEC revival is the other interesting data point. Reviving a state generator after three decades of privatisation isn’t a trivial administrative act. It’s an admission that the market, left alone, wasn’t going to deliver the pace of firmed renewable capacity the state wanted on the timeline it wanted. Whether the SEC can actually execute at the scale promised is still an open question, particularly against the offshore wind feasibility grind in Gippsland, but the intent tells you something about where state governments’ confidence in pure market delivery has landed.

The downside of public ownership is equally real and shouldn’t be waved away. Government-owned projects answer to budget cycles, not just capital markets, and that means cost overruns get absorbed by taxpayers rather than shareholders, with far less market discipline forcing a rethink. HumeLink’s cost nearly tripling is a Transgrid-adjacent example of exactly that dynamic, even though Transgrid itself sits in a hybrid ownership structure. When the project is public, there’s rarely a moment where the whole thing gets cancelled because the numbers no longer stack up. That can be a feature. It can also be how you end up fifty percent over budget with nobody accountable to a share price.

Where the two models actually meet #

The cleanest way to see this is through firming capacity, because it’s the part of the grid where both models are being tested side by side right now. Gas peakers versus big batteries is largely a private-capital argument, fought out between gentailers and battery developers chasing the same contracts for difference and capacity mechanisms. Pumped hydro at the scale Queensland and Tasmania are attempting is almost entirely a public-capital argument, because the payback horizon simply doesn’t suit private discount rates.

Tasmania’s Battery of the Nation concept depends on Marinus Link, and Marinus is instructive precisely because it keeps slipping between the two models: part state-owned, part federally backed, never quite purely either. That ambiguity has arguably slowed it down. Projects with unclear ownership lines tend to attract unclear accountability, and Marinus has had both cost staging revisions and timeline questions that a purely private or purely public structure might have resolved faster, one way or the other.

Western Australia’s isolated SWIS grid offers a cleaner comparison because there’s no interconnector complicating the picture. Synergy, the state-owned generator-retailer, is building out the Collie battery alongside privately-backed Neoen capacity in the same system, on the same transition timeline, replacing the same retiring coal. It’s about as close to a controlled experiment as the NEM-adjacent world gets, and early indications are that neither model is dramatically outpacing the other on construction timelines. The difference shows up more in who absorbs the risk when something goes wrong.

What the numbers actually say #

AEMO’s Integrated System Plan doesn’t pick a side on ownership, and it shouldn’t, because its job is modelling the capacity Australia needs, not who should own it. But the ISP’s own cost estimates for the transmission and firming build-out, running past $20 billion in some iterations, make clear that no single ownership model is going to deliver all of it. The scale is too large and too varied. Some of it wants private discipline. Some of it wants patient, low-cost public capital that doesn’t need to clear a commercial hurdle rate within a decade.

The Clean Energy Regulator’s Clean Energy Regulator data on committed and probable generation projects shows the private pipeline is still doing most of the heavy lifting on wind and solar capacity additions, which makes sense given those technologies have reasonably well-understood economics and short build times. The long-duration storage and network side is where public money is doing disproportionately more of the committed capital, which also makes sense given the payback horizons involved.

I’d argue the mistake both sides make in this argument is treating it as a binary when the actual answer is sectoral. Private capital is well suited to assets with predictable, bankable revenue streams over ten to fifteen years. Public capital is better suited to assets that earn out over twenty-five to forty years, or that carry construction risk too large for any private balance sheet to wear alone. Pretending one model should dominate the whole build-out is where most of the political posturing on this subject goes wrong.

The verdict #

So who builds it better? Neither, cleanly, and that’s not a dodge, it’s the finding. Private ownership has built the bulk of Australia’s wind and solar fleet faster and more cheaply than a state-run equivalent likely would have, and the AGL-Origin rivalry has kept retail pricing more contestable than a monopoly utility ever would. But private capital has consistently under-delivered on the assets the grid most desperately needs right now, long-duration firming and interstate transmission, because the payback math doesn’t clear a private hurdle rate inside a politically tolerable timeframe.

My read, for what it’s worth after watching Borumba’s budget swell and HumeLink’s cost nearly triple in the same two years Tilt Renewables quietly finished three projects on schedule: public ownership should own the long-duration, long-payback assets, and private capital should keep building everything else. Treat it as division of labour, not ideology. The states that work that out fastest, rather than re-litigating privatisation as a culture war, are the ones that will hit their 2030 targets with the least public money wasted. Like a bowling attack that needs both a strike bowler and someone who can hold an end for thirty overs, the grid needs both kinds of owner. Pretending one alone can do the whole job is how you end up with neither doing its job well.

– Marcus Wren, Editor

Photo by Matthew Henry on Unsplash