The electricity load nobody is talking about
AI data centres are the fastest-growing new electricity load in Australia. Not the fastest-growing industrial load. The fastest-growing load, full stop.
A single large-scale AI training facility draws between 100 and 500 megawatts continuously. That is roughly the output of a mid-sized gas peaker plant, running flat, around the clock, every day of the year. The inference workloads that serve commercial AI products run permanently on top of that. As Australian businesses adopt AI tools, as government agencies move procurement onto AI-assisted platforms, and as the global hyperscalers build out their Asia-Pacific capacity, that load will compound.
The International Energy Agency has documented the trajectory globally. In Australia, the Australian Energy Market Operator has flagged data centre demand as a material variable in its grid forecasting. This is not a projection borrowed from overseas. It is already in the domestic planning numbers.
The question that follows is simple: who pays for it, and who captures the value?
What the living cost indexes actually show
Before getting to the opportunity, the cost.
The ABS Selected Living Cost Indexes, which measure cost changes for specific household types and are not the same as the CPI (they include mortgage interest charges, which the CPI excludes, so the two must never be quoted interchangeably), show what has happened to Australian households since 2012.
From the first quarter of 2012 to the second quarter of 2026:
- Other government transfer recipient households: +51.7%
- Pensioner and beneficiary households: +50.0%
- Age pensioner households: +48.5%
These are index changes for household types, not what any individual household experienced. But the direction is unambiguous, and the households at the top of that list are not the ones best placed to absorb another round of energy price increases driven by new industrial demand on the grid.
Electricity is embedded in every one of those figures. It is in the food supply chain, in transport, in heating and cooling, and now increasingly in the digital services that households and small businesses depend on. When a new and permanent electricity load the size of several industrial precincts arrives on the grid without a corresponding increase in generation capacity, the cost lands somewhere. It lands on the bill.
The Nationals understand this mechanism. Their cost-of-living platform is built around the argument that energy prices are not a side issue: they are the transmission channel through which resource and infrastructure decisions reach the household. That framing is correct, and it connects to the AI infrastructure question in a way that most of the commentary has missed.
The competitive lever
Australia holds something that almost no other developed economy holds in the same combination: abundant renewable energy potential, existing fossil fuel export infrastructure, a stable regulatory environment, and a resource base that generates enough export revenue to fund the transition without borrowing against future generations.
Cheap, reliable, domestically-owned power is what makes Australian-owned compute viable. This is not an abstract claim about national pride. It is an industrial cost argument.
AI compute is electricity-intensive in a way that few industries are. The operating cost of a data centre is dominated by power. A jurisdiction that can deliver power at a structurally lower cost than its competitors can host compute at a lower cost than its competitors. That is the lever.
Norway built a sovereign wealth fund worth roughly US$1.8 trillion for 5.5 million people from its oil and gas revenues. Australia exports more resource value per capita and has no equivalent fund. The comparison is not made to shame anyone. It is made because the mechanism that Norway used, capturing resource value domestically rather than exporting it raw and importing the finished product, is exactly the mechanism available to Australia on compute.
The alternative is the current trajectory: Australian businesses and government agencies pay for AI services delivered from data centres in the United States, Singapore or the Gulf, running on power generated elsewhere, owned by entities incorporated elsewhere, with the economic surplus flowing elsewhere. The power bill is the policy.
What the Nationals are actually arguing
The Nationals' cost-of-living platform centres on energy affordability, regional infrastructure and domestic resource use. Read through the lens of the compute build-out, those three things are not separate issues.
Energy affordability for households and energy affordability for data centres are the same problem approached from different ends. A grid that is cheap and reliable enough to attract large-scale compute investment is a grid that is also cheaper for the pensioner household that showed up in the ABS living cost indexes above. The investment in generation capacity that makes one possible makes the other more likely.
Regional infrastructure matters because the best sites for large-scale renewable-powered data centres are not in the Sydney CBD. They are in regional areas with land, water access, renewable resources and existing grid connections. The Nationals represent those areas. They have a direct constituency interest in making the case that regional Australia is not just a resource extraction zone but a location for value-adding industry, including digital infrastructure.
Domestic resource use is the third leg. The argument that Australia should capture more value from its resource base before exporting it is the same argument applied to compute: build the infrastructure here, run it on domestic power, keep the economic surplus onshore.
None of this requires a detailed AI policy. The Nationals do not currently have one, and that is a gap worth naming. But the energy and resources platform they do have is the correct foundation for one, and it connects to the cost-of-living argument in a way that the current government's framing does not.
The Labor gap
Labor's response to cost-of-living pressure has been framed primarily around cost-of-living payments and energy bill relief. Those are real interventions and they show up in household budgets. But they do not address the structural question of why the grid is under pressure, who is adding to that pressure, and whether the value being generated by the new load is staying in Australia.
The fastest-growing new electricity load in the country is AI data centres. The government has approved foreign investment in data centre infrastructure. The AI procurement spend of Commonwealth agencies flows substantially to offshore vendors. AusTender publishes no usable feed that would allow the offshore share of AI contract value to be calculated, so the precise figure is not available and this piece will not invent one. But the mechanism is documented: Australian public money, spent on AI services, delivered from infrastructure that may be foreign-owned, running on Australian power, with the surplus leaving the country.
That is a cost-of-living argument. It is also a sovereignty argument. The Nationals are better positioned to make it than any other party in the current parliament, because they represent the regions that hold the energy and the resources, and they have a constituency that understands what it means when the value of what you produce leaves without coming back.
The policy gap that needs filling
The Nationals' cost-of-living platform does not yet connect the energy argument to the compute argument explicitly. That connection needs to be made, and made in policy terms: what planning rules govern data centre siting in regional areas, what grid investment is required to support large-scale compute loads, what conditions attach to foreign investment approvals for digital infrastructure, and what domestic ownership requirements, if any, apply to AI infrastructure that processes government data.
These are not abstract questions for a future parliament. The investment decisions are being made now. Data centres have 20 to 30-year asset lives. The ownership structure and the power contracts that are signed in the next three years will determine whether Australia captures the value of the AI build-out or pays for it from the other side of the ledger.
The Nationals have the right platform to anchor that argument. The energy and resources base is the competitive lever. Cheap domestic power is what makes Australian-owned compute viable. The cost-of-living relief that households need and the industrial policy that keeps the surplus onshore are the same argument, approached from different ends.
That argument deserves a policy to go with it.