MATTERS OF PUBLIC IMPORTANCE

Ms STEGGALL (Warringah) (15:31): Thank you to the member for Warringah for bringing forward this matter of public importance to the parliament. The National AI Plan commits to capturing the opportunity of AI, spreading the benefits and keeping Australians safe. My concern today is whether we really have the policy settings to achieve all three, or whether we are trading long-term sovereignty for a short-term sugar hit.

Let's be honest: the opportunity is real. Australia has genuine competitive advantages, abundant renewable energy, land, geopolitical stability and proximity to the Indo-Pacific market. In 2024, we ranked second globally as the most attractive destination for data centre investment.

Data centre investment has kept GDP out of negative territory, so this really matters, and AI data centres have been and will be an important part of our economy. And be honest: megabytes are far easier to export than hydrogen. Unlike our earlier superpower ambitions, digital infrastructure doesn't need a port, but it still needs power, and there is a strong sovereign capability case for having AI training and inferencing happening here on Australian soil.

This case was made sharply when the US issued an export-control directive to Anthropic, cutting off Australia's access to frontier AI models overnight. Australia already has more than 160 operational data centres, and there are more than 90 in the pipeline. In New South Wales alone, there are 44 projects totalling 11.4 gigawatts, equivalent to nearly four Eraring coal stations.

It's a dominant driver of new electricity demand, and there are many people concerned that we can have the clean energy transition, or we can have data centres, but we can't have both. But we need to have both. AI has voracious appetite for land, energy, water and construction.

Other countries have acted. Singapore imposed a moratorium. The US had tech companies sign a ratepayer protection pledge.

Ireland now requires data centres to source 80 per cent of demand from new renewable energy. The public is right to be concerned. No-one wants big tech put before the planet or corporations put before Australians.

Trust that the government will defend Australia's interests is fraying. Let's start with energy consumption. I recognise that the government has put in a statement of expectations and that this sends an important signal.

I note the minister's comments that this is partly because of the different jurisdictional challenges in Australia and the roles of local, state and federal government. But businesses need certainty, and so do Australians. The work that the government is doing in this space needs to speed up so that we have greater certainty about what we are going to get as a nation when these data centres are being put in every single day.

The point that we also need to make is that the broader energy framework is not coherent, and that is a problem too for data centres. Centralised data centres must bring their own renewable energy under the statement of expectations, but the same compute distributed across office buildings carries no equivalent obligation, despite being less efficient. The treatment of new versus existing industrial demand is equally arbitrary when it comes to electricity.

Tomago Aluminium draws close to a gigawatt—roughly 10 per cent of New South Wales electricity supply—and not only faces no BYO requirement of renewable energy but will actually have its energy use subsidised. There may be good reasons for these distinctions, but the government hasn't articulated them, and we do need a coherent approach in terms of industrial-scale energy requirements.

The deeper problem is that the government is approving demand that, at the moment, we cannot supply. New South Wales solar projects now average 1,384 days to clear planning—almost double 2023 levels and more than 10 times Western Australia's 94-day average. Data centres can be built far faster than the renewable energy to power them.

That temporal mismatch is where the anxiety lies. If it isn't closed, Australians will pay through higher emissions, higher prices or both. Unlike approval, global demand for compute cannot be paused.

Refuse the investment and it simply moves beyond Australia's reach. The answer is faster infrastructure approvals, locational price signals that guide investment to where the grid really can support it and the discipline that supply must precede demand, not chase it. Australia has been here before.

We've extracted gas, exported it and watched Australian industry pay some of the highest gas prices in the world for a domestic resource when we had expanded supply. We must not make the same mistake with compute. Australian energy should benefit Australians.

That means energy abundance, with industry that follows, not competes with, households. It means sharing the productivity gains from the technology and building domestic sovereignty. We can't hold back AI; we need to build with it.