On a stretch of Mamre Road in Western Sydney, plans are being drawn for one of the biggest data centres on the planet: six four-storey buildings and more than 900 cooling units. If built to full capacity, it would draw more power than the Tomago aluminium smelter, making a single data centre Australia’s largest energy user.
That is the scale of what’s coming. And according to Peter Cardassis, technical services director at Logicalis Asia Pacific, it’s exactly the conversation Australia isn’t having.
“Australia is having the wrong conversation about AI, and it risks missing the next phase of global advantage because of it,” he says. “Our national debate is focused on policy frameworks, ethics, guardrails, and access to compute. Those issues matter; however, they are not the real constraint on Australia’s AI future. The real bottleneck is sustainable infrastructure.”
The boom is already here
Whatever you think of that argument, the money agrees with him. Australia now hosts more than 250 data centres and ranks among the most attractive destinations in the world for data centre investment. Amazon has committed $20 billion to expanding its Australian footprint, Microsoft $5 billion, and investment giant Blackstone paid $24 billion to acquire Sydney-born operator AirTrunk. Analyst firm Gartner expects Australian data centre electricity consumption to jump almost 38 per cent this year alone, driven overwhelmingly by AI workloads.
Every ChatGPT query, every Canva render, every AI feature quietly added to the apps we use daily runs on this physical machinery, power, land, water and cooling. AI may feel like software, but it is built out of infrastructure.
Which is why Cardassis argues the ground is shifting under the boardroom conversation. “ESG is no longer a reputational issue,” he says. “It has become a factor in AI investment decisions. Organisations want AI capability. Yet they also want to know how it will be powered, how sustainable it is, and whether it aligns with long-term environmental commitments.”
Australia’s overlooked advantage
Here is where his argument gets interesting for anyone who grew up hearing that Australia’s economy rides on what we dig out of the ground. The inputs that decide where the world’s AI gets built, abundant renewable energy, available land, water, political stability, happen to be things Australia has in rare combination.
“Few countries have the combination of renewable energy potential, available land, and political stability required to scale AI sustainably,” Cardassis says. “Australia has an abundance of solar, wind, and open space. If we design our infrastructure correctly, these assets could underpin a new category of digital infrastructure: sustainable sovereign AI infrastructure.”
The prize, as he sees it, is hosting the workloads that power the global AI economy and attracting hyperscaler capital and the regulated industries, from healthcare to financial services, that need trusted places to run sensitive AI systems.
“Australia would not simply adopt AI technologies developed elsewhere; we would help power them for the world economy. This becomes another Australian export and, in the future, may be more valuable to our country than the mining industry.”
A claim worth interrogating
Bigger than mining is a big call — iron ore alone earns Australia over $100 billion a year and the road there is not guaranteed. The Climate Council counts around 90 more data centres in the national pipeline, but grid connections, water demand and community consent all stand between announcement and operation. Analysis by the United States Studies Centre warns that without new renewable generation and storage, data centre growth could push wholesale electricity prices up as much as 26 per cent in NSW by 2035 which is a cost every household would feel.
In other words: the opportunity is real, and so is the risk of doing it badly.
Canberra appears to have noticed. Following the National AI Plan released in late 2025, the federal government in March issued formal expectations for data centre and AI infrastructure developers, tying regulatory priority to contributions on clean energy, water security and local jobs. It’s an early sign the national conversation is moving in the direction Cardassis is pushing from AI as a software debate to AI as nation-building infrastructure.
His prescription goes further: treat sustainability metrics in the National AI Plan not as compliance boxes, but as the basis of a global competitive edge. “Integrating renewable energy, storage, and modern grid capacity into AI expansion plans will ensure growth is sustainable rather than reactive,” he says.
What it means for the rest of us
For business leaders, the takeaway is that AI strategy can no longer stop at which tools to roll out. Energy demand, cooling requirements and ESG accountability will increasingly shape which AI plans are viable at all — and the companies that scale AI successfully will be those that treat infrastructure and sustainability as core strategy, not afterthoughts.
For the rest of us, it reframes something bigger. The AI era’s most valuable Australian export may not be an app, a model or a unicorn startup — it may be the clean electricity, land and engineering that keep the world’s AI running. That would create demand for skills well beyond coding: renewable energy, grid engineering, sustainability strategy, construction, policy.
“Australia now has a choice,” Cardassis says. “We can continue to treat AI as primarily a software conversation. Or we can recognise that the next phase of the AI economy will be defined by infrastructure, energy, land, cooling, and sustainability, and plan accordingly.”
The wrong conversation, it turns out, might be the one we’re not having at all.
FAQ: Australia’s AI infrastructure opportunity
What is “sustainable sovereign AI infrastructure”? Data centres and AI compute capacity located in Australia, powered predominantly by renewable energy, and operated under Australian law — allowing sensitive workloads from government, healthcare and finance to run onshore.
Why is Australia attractive for AI data centres? A rare combination of world-class solar and wind resources, available land, water access relative to many regions, political and regulatory stability, and proximity to Asia.
How much energy do Australian data centres use? Gartner forecasts Australian data centre electricity consumption will rise about 38 per cent in 2026 to 6.2 terawatt hours, more than doubling again by 2030.
What is the Australian Government doing about it? Under the National AI Plan (December 2025), the government released formal expectations for data centre and AI infrastructure developers in March 2026, covering clean energy, water security, workforce and community benefit.
This article draws on commentary supplied by Logicalis Asia Pacific, with additional reporting.
