What Happened
On the corner of Mamre Road and the yet-to-be-built South Link Road, in the western Sydney suburb of Kemps Creek, plans are afoot that could change the way the world works. It is here, in this semi-rural expanse wedged between Sydney's rapidly growing outer suburbs and the new international airport at Badgerys Creek, that the AI boom has landed in Australia, with a thud.
If the site owners, IFM Investors, can overcome concerns of local residents, the 52-hectare site will house one of the world's biggest data centres, to be built and operated by US-owned Airtrunk at a cost of more than A$5 billion. Incorporating six four-storey buildings, 936 cooling units, 852 diesel-powered back-up generators and a huge diesel storage facility, Airtrunk is hoping to cash in on the rush to AI and the potential role Australia could play as a regional powerhouse.
Microsoft's Satya Nadella visited in April, pledging A$25 billion in data centre investment, while OpenAI boss Sam Altman appeared via video link at a conference organised by the Australian Financial Review. The numbers are genuinely large: more than A$150 billion is lined up, waiting to be poured into this frenzied race to build AI infrastructure, prompting some analysts to invoke memories of the mining boom. Almost 20 per cent of all non-residential construction is now being directed towards building these facilities.
Why It Matters
The scale of the investment pipeline is striking on paper. "Alongside the potential A$200 billion energy transition investment pipeline, Australia is seeing an investment boom, with the total pipeline approaching 13 per cent of GDP, rivalling the size of the mining investment boom," according to analysis cited by ABC News.
But the mining boom comparison cuts both ways. "Compared to Australia's early 2010s mining capex boom, which peaked at nearly 7 per cent of GDP, the tech dynamic appears much smaller so far," one analyst wrote. The mining boom delivered massive royalty streams, local employment, and supply-chain development. The data centre boom's domestic economic footprint is less certain, partly because "a meaningful share of data centre fit-out relies on imported capital equipment."
There is also a tax dimension that deserves scrutiny. ABC News reported concerns about transfer pricing arrangements, with some operators accused of minimising Australian tax obligations "purportedly on the basis that the Australian data centre business is merely providing a service to its foreign associates." That framing - if accurate - would mean Australia hosts the physical infrastructure, absorbs the energy costs and planning risk, and sees a reduced share of the taxable profit.
For Australian businesses thinking through their own AI automations roadmap, the macro picture matters: the terms on which this infrastructure is built will shape the cost and availability of AI services for years.
Key Details
The energy question is the most immediate pressure point. The proposed Mamre Road development in western Sydney will use about 25 per cent more power than the Tomago aluminium smelter north of Newcastle - one of the largest single industrial power consumers in the country. Scale that across the full pipeline and the grid implications become serious.
"If data centre growth is not matched with new renewable generation and storage, this could increase wholesale prices by more than 20 per cent across our main grid by 2035, on average - and up to 26 per cent in NSW and 23 per cent in Victoria," according to modelling cited in the ABC report. That is not a distant hypothetical; it is a planning constraint that affects every business paying an electricity bill in the two most populous states.
On the jobs side, the picture is more nuanced than either boosters or critics suggest. "The state of AI, and quite frankly even for the foreseeable future, is more about what I'll call task level automation inside of jobs," one expert noted - meaning the workforce disruption from AI is more likely to reshape roles than eliminate them wholesale in the near term. That framing has direct implications for AI training investment inside Australian organisations.
About A$13 billion was invested into the data centre sector in a recent period, helping deliver a record 16 per cent rise in machinery and equipment investment - a result that took many by surprise. The question is how much of that machinery was sourced domestically.
Background and Context
Australia's position in this race is genuinely competitive on fundamentals. "Australia has among the best natural resources and abundant clean energy [stores] in the world, and if Australia wanted to become a data centre capital of the world, it would certainly be able to," one industry figure said. The combination of political stability, strong rule of law, proximity to Asian markets, and renewable energy potential makes the country a credible regional hub candidate.
The federal government has moved to formalise its expectations. The Department of Industry, Science and Resources published guidance titled "Expectations for data centres and AI infrastructure developers," signalling that Canberra wants commitments on local jobs, local procurement, and community benefit - not just headline investment figures. That document represents the government's attempt to avoid the situation where Australia hosts the hardware but exports the value.
The AI strategy challenge for Australian policymakers mirrors the challenge for Australian businesses: capturing genuine value from AI adoption requires deliberate planning, not passive participation. Organisations that treat AI as something happening to them - rather than something they are actively shaping - tend to end up on the wrong side of the ledger. Our work across professional services and other sectors consistently shows that early, structured engagement with AI produces better outcomes than reactive adoption.
What Comes Next
The Kemps Creek development is a bellwether. If IFM Investors and Airtrunk navigate the local planning concerns and the project proceeds, it will set a template - on energy agreements, tax treatment, local employment ratios, and community obligations - that other developers will reference. The federal government's published expectations give advocates and regulators a baseline to hold operators against.
The energy investment question is the critical variable. If the renewable generation pipeline keeps pace with data centre demand, Australia avoids the worst-case electricity price scenarios and potentially builds a genuine competitive advantage in low-carbon AI compute. If it does not, the country risks subsidising foreign operators' energy costs through higher prices for everyone else.
For Australian businesses, the practical implication is straightforward: the AI infrastructure is being built regardless of how the policy debates resolve. The organisations that will benefit most are those building internal capability now - understanding AI-driven SEO and GEO, deploying automations that reduce operational costs, and developing the strategic clarity to make good decisions as the technology matures. Waiting for the dust to settle is itself a choice, and not a neutral one.
Frequently Asked Questions
What is the total scale of AI data centre investment lined up for Australia?
More than A$150 billion is lined up for AI infrastructure investment in Australia, according to reporting by ABC News. When combined with the energy transition investment pipeline, the total approaches 13 per cent of GDP - a figure that analysts are comparing to the scale of the early 2010s mining boom, though with important caveats about how much of that value will remain in the Australian economy versus flowing back to foreign parent companies.
Why is electricity pricing a concern for Australian households and businesses?
Data centres are extraordinarily power-hungry facilities. The proposed Kemps Creek development alone would consume about 25 per cent more power than the Tomago aluminium smelter, one of Australia's largest industrial electricity users. Modelling cited by ABC News found that if data centre growth is not matched with new renewable generation and storage, wholesale electricity prices could rise by more than 20 per cent nationally by 2035, with NSW facing increases of up to 26 per cent and Victoria up to 23 per cent. Those costs flow through to every business and household connected to the main grid.
What is the Australian government doing to ensure the country benefits from the boom?
The Department of Industry, Science and Resources has published formal expectations for data centre and AI infrastructure developers, outlining what the government wants to see in terms of local jobs, local procurement, and community benefit. This document gives regulators and advocates a reference point when assessing whether individual projects are delivering genuine domestic value or simply using Australia as a low-risk location to host foreign-owned infrastructure.
How does the data centre boom compare to the mining boom in terms of local economic benefit?
The comparison is instructive but imperfect. The mining boom peaked at nearly 7 per cent of GDP in capital expenditure terms and generated substantial royalty revenue, local employment, and supply-chain development. The data centre pipeline is larger in headline terms but a meaningful share of the fit-out relies on imported capital equipment, and transfer pricing arrangements may reduce the Australian tax take. The domestic multiplier effect is likely to be smaller per dollar invested than the mining boom delivered, though the full picture will depend on how policy settings evolve.
What does this mean for Australian businesses adopting AI?
The infrastructure build-out will make AI compute more accessible in Australia over the coming years, which is broadly positive for businesses looking to adopt AI tools and automation. However, the energy cost risk is real and could affect operating costs across the economy. The more immediate implication is that organisations which invest now in building genuine AI capability - through structured strategy, staff training, and process automation - will be better positioned to extract value from the infrastructure as it comes online, rather than scrambling to catch up once the market has moved.
