Australia’s race to build the physical backbone of artificial intelligence has collided with one of the country’s most contested energy resources. A proposal to power a $28 billion AI data centre using shale gas drawn from the Northern Territory has landed squarely in the middle of a national argument about how the enormous electricity appetite of AI should be fed, and at what cost to the climate.
The plan, reported by World Oil, would see gas from the Beetaloo Basin, a vast shale formation south of Katherine, used to generate the round-the-clock electricity that a hyperscale computing campus requires. It is one of the clearest signals yet that Australia’s gas industry sees AI infrastructure as a new anchor customer at a moment when traditional export markets are maturing and the domestic conversation has turned firmly towards renewables.
Why gas and AI have found each other
The logic behind pairing shale gas with a data centre is not hard to follow. Large AI training and inference facilities draw power in the hundreds of megawatts and expect it to be available every hour of every day, regardless of weather. That demand profile has made “firm” generation, the kind that can run continuously and ramp on command, unusually valuable. Around the world, developers chasing AI capacity have increasingly turned to gas, and in some cases nuclear and coal, because the grid connections and firmed renewable supply they would prefer simply cannot be built fast enough.
For the Beetaloo, a basin whose commercial future has always depended on finding buyers close enough to justify the cost of moving its gas, a power-hungry data centre offers something rare: a large, local load that consumes the resource where it is produced rather than shipping it thousands of kilometres. That changes the economics of a region that has struggled for years to prove up a viable path to market.
The Beetaloo has long been touted by governments and industry as a potential engine of northern development. It has also been a lightning rod. Traditional owners, environmental groups and climate scientists have fought its expansion for the better part of a decade, arguing that opening a major new gas province is incompatible with Australia’s emissions commitments. Attaching that gas to the fast-growing, publicly celebrated AI sector does not make those objections disappear. If anything, it sharpens them.
Two ways of reading the same proposal
Supporters frame the project as pragmatic nation-building. On this view, the AI boom is going to happen whether or not Australia participates, and the data centres underpinning it will be built somewhere. Better, the argument runs, to host that capacity onshore using domestic energy, capturing the jobs, the tax base and the sovereign computing capability, than to watch the workloads and the economic value flow to the United States or the Middle East. Gas, in that telling, is a bridge that gets the compute built now, with the option to green the supply later.
Critics see the same facts and reach the opposite conclusion. They point out that a data centre running continuously on fossil gas locks in a heavy emissions profile for the life of the asset, which can run to two decades or more. They question whether “bridge” fuels ever get replaced once the capital is sunk, and they note that the AI industry’s own sustainability pledges sit awkwardly beside a plan to burn newly extracted shale gas. There is also the water question. Data centres consume large volumes for cooling, and the arid Top End is not an obvious place to site a thirsty facility, a tension FluentSea has explored in its coverage of data-centre cooling efficiency.
What it means for Australia
This is not an abstract international story. It goes to the heart of a policy tension Australia is living through right now. The Commonwealth has been working through how to reconcile surging data-centre demand with its renewable-energy targets, an effort captured in the developing framework for powering AI data centres. A high-profile proposal to run one of the largest such facilities on shale gas is a direct test of whether that framework can hold, or whether the sheer speed of AI investment will simply route around it.
The financial stakes are also real. Australia has quickly become a magnet for AI infrastructure capital, with Firmus and its backers committing billions to build so-called AI factories, and companies such as Maas Group winning major construction contracts on the back of that pipeline. A $28 billion figure, if it holds up, would rank among the largest single infrastructure commitments the country has seen in years. Where and how that money is spent will shape regional economies, the grid and the emissions ledger for a generation.
There is a geographic dimension too. Much of Australia’s data-centre growth has clustered in Sydney and Melbourne, close to fibre, users and existing power. A Northern Territory project powered by local gas would push serious compute capacity into the north, near Darwin’s ports and defence infrastructure and closer to Asian markets. That could be a genuine boon for a jurisdiction long promised a resources-led future that has been slow to arrive, provided the environmental and social licence questions can be answered.
What happens next
A headline number and a concept are not a built data centre. Projects of this scale face a long road through environmental approvals, gas-supply agreements, grid or off-grid power arrangements, water access and, crucially, a signed anchor tenant willing to house workloads there. The Beetaloo’s own development timeline remains uncertain, and any facility depending on its gas inherits that uncertainty.
The bigger contest is over precedent. If a shale-gas-powered AI campus wins approval and financing in the Territory, it will embolden others to pitch fossil-fuelled compute elsewhere in the country. If it stalls on climate, cost or community opposition, it will reinforce the push to tie Australia’s AI infrastructure to firmed renewables and storage instead. Either way, the proposal has made concrete a question the nation can no longer defer: what, exactly, are we prepared to burn to run the machines we are being told will define the coming decade.
Sources: World Oil.
