The $16 billion battery that runs on electricity it doesn’t have
- Grant McLachlan
- Mar 30
- 6 min read
There is a particular kind of infrastructure madness that only becomes visible once you draw a map.
Lake Onslow is that madness, now officially fast-tracked — meaning Infrastructure Minister Chris Bishop has decided that New Zealand’s next great energy project should be a scheme that consumes electricity in order to produce electricity, at a cost that last reached $15.7 billion before the government of the day quietly dropped it, at a location that requires inundating 7,100 hectares of Central Otago hill country, and whose backers concede it is a good ten years from generating a single watt.
Someone in Wellington thinks this is progress.
The scheme works like this: when electricity is cheap and plentiful, water is pumped uphill from the Clutha River into an enormously enlarged Lake Onslow, sitting 769 metres above sea level. When demand rises and prices spike, that water is released back down through turbines into the valley below. The consortium now championing the project — which includes former Labour minister David Parker, former Meridian Energy chief executive Keith Turner, and Reserve Bank chair Rodger Finlay — describes it as “an absolute game changer.”
What he does not say — what the entire pumped hydro conversation politely avoids — is that a battery does not generate electricity. It stores it. And storing electricity through pumped hydro costs you roughly 20 to 25 cents in every dollar, because the physics of pumping water uphill and recovering it on the way down returns only 75 to 80 percent of what you put in. Lake Onslow will be a permanent net consumer of electricity. Before it can release a single megawatt, New Zealand must first produce the electricity to charge it. In a country that already relies on burning coal at Huntly whenever the hydro lakes run low, that is not a trivial condition.
The question the proponents never answer is: Where does the charging electricity come from?
I have been writing about New Zealand’s electricity market for decades. What I have documented, repeatedly, is an industry structurally incentivised to withhold generation capacity — because scarcity drives prices, and high prices are extraordinarily profitable for the four gentailers who dominate generation and retail simultaneously. As I showed in ‘Rivers of Wasted Power’, the combined recoverable gap across the Waikato, Waitaki, and Clutha systems is approximately 3,100 GWh per year: enough to eliminate coal generation in a drought year, from water already flowing through dams we built and paid for.
Lake Onslow does not solve that problem. It decorates it. A giant battery controlled by the same market participants who benefit from price volatility is not a public good. It is an amplifier.
The entrenched pattern here is not difficult to trace. New Zealand’s electricity market was designed, in the 1990s, to break a state monopoly and create competition. What it actually created was a private oligopoly with far less accountability than the entity it replaced. The four gentailers who now dominate generation and retail simultaneously collect a combined operating profit forecast at $1.86 billion in the six months to December 2025 — a 44 to 45 percent increase on the same period the year before — while roughly 30 percent of New Zealand households face energy hardship. The competition laws that should discipline this market are toothless by design. And Lake Onslow, a decade away and costing billions, poses no immediate threat to that arrangement whatsoever.
Which brings me to a question I have not seen asked once in the coverage of this fast-track decision: What is sitting between Lake Hawea and Lake Dunstan?
Lake Hawea’s operating level runs between 338 and 346 metres above sea level. Lake Dunstan operates between 193.5 and 194.5 metres above sea level. That is a hydraulic head of approximately 144 to 152 metres — 2.5 times the 59-metre head that drives Ōhau A, and greater than the combined head across any two of the three Ōhau stations. The entire 148-metre drop is available through a single canal and powerhouse, rather than a cascade of three. The Hawea River flows from the lake’s southern outlet near Luggate into the Clutha/Mata-Au, which descends through Cromwell into Lake Dunstan. The geography is not merely similar to the Ōhau A, B and C canal scheme in the Mackenzie Basin. It is the same geometry, on a larger scale.
Ōhau A, fed by a manmade canal carrying water from Lakes Ōhau and Pūkaki, achieves 264 megawatts from four generating units — enough electricity for around 144,000 average New Zealand homes. Ōhau B and C are twin stations, each producing up to 212 megawatts, built by running the water progressively down through the Waitaki cascade. The model is proven, efficient, and already operating at three sites in the South Island. A canal and powerhouse between Luggate and Lake Dunstan, capturing the 148-metre head available from Lake Hawea, could at maximum consented flows of 200 cumecs generate in the vicinity of 200 to 260 megawatts — roughly 1,400 to 1,940 GWh annually — continuous, renewable, requiring no electricity to charge, and drawing on storage that already exists.
The numbers warrant spelling out. Ōhau A produces 1,150 GWh per year; Ōhau B and Ōhau C generate 970 GWh each — combined, the three stations deliver around 3,090 GWh annually from a mean Waitaki system flow of approximately 364 cumecs. The Clutha at Cromwell carries a mean flow of around 500 cumecs — roughly 38 percent more water than flows through the Ōhau stations. A Luggate canal, drawing on Lake Hawea’s 148-metre head and consented flows of up to 200 cumecs, would generate in the vicinity of 1,400 to 1,940 GWh annually from a single station — exceeding Ōhau A’s 1,150 GWh output, and at the higher end comparable to Ōhau A and Ōhau B combined. The modest 17-megawatt generator Contact Energy consented at the Lake Hawea dam outlet — and never built — would have produced roughly 90 to 100 GWh per year.
Lake Onslow, by contrast, will produce zero GWh net annually. Its backers cite 5,000 GWh of storage capacity and 1,000 megawatts of installed generation capacity. What they do not cite is net annual output, because there is none. Every gigawatt-hour Lake Onslow discharges must first be purchased from the grid and pumped uphill, with 20 to 25 percent permanently lost to physics.
In energy accounting terms, Lake Onslow generates nothing. It launders electricity at a structural loss.
Contact Energy was actually granted resource consents over a decade ago to install turbine generators at the Lake Hawea dam — a scheme capable of producing around 17 megawatts — and simply never built it. The company is now, separately, seeking to lower Lake Hawea’s operating level by two metres through the same fast-track process being used for Lake Onslow. We are fast-tracking the management of scarcity, not the creation of abundance.
Nobody is asking why.
That inaction is the story. Not the resource consent that wasn’t exercised, but the market conditions that made inaction more profitable than generation.
New Zealand’s electricity oligopoly does not want abundant cheap generation. Abundance destroys the scarcity premium on which their financial models depend.
Lake Onslow — a decade away, costing billions, dependent on surplus electricity that doesn’t reliably exist — poses no such threat. A canal from Hawea to Dunstan, producing clean, new, cost-effective generation from existing stored water and existing hydraulic head, would.
That is why one is being fast-tracked and the other has never been built.
The electricity industry in this country has spent thirty years perfecting the art of visible inaction dressed as complexity. Lake Onslow is the masterpiece. A battery the size of a district, charged with electricity we don’t have, built by the same interests that profit from the problem it purports to solve, fast-tracked by a minister who seems not to have asked why Central Otago’s rivers already run past one of the most compelling untapped generation sites in the South Island.
Draw the map. Then ask the question.
