Under a cloud: the growing resentment against the massive datacentres sprouting across Australian cities | Computing

The Rise of the Hyperscale AI Factory

The digital landscape is shifting from traditional cloud storage to what industry leaders call hyperscale AI factories. These aren’t your standard server rooms; they are massive industrial complexes designed for accelerated computing, capable of powering the generative AI tools that are reshaping global productivity.

In Australia, this boom is manifesting in projects like the M3 datacentre in West Footscray. If fast-track planning is granted, this facility is projected to cover 10 hectares and draw 225MW of power, running 24/7 to meet the relentless appetite of AI processing.

However, the scale of these “factories” is creating a friction point. As these sites expand, they often collide with residential boundaries, turning quiet suburbs into the front lines of a technological arms race.

The Battle for the Backyard: Why Zoning is Failing

A recurring theme in the expansion of AI infrastructure is the failure of legacy zoning. Many of these facilities are built on land zoned as “industrial,” but as these sites scale up, the distinction between an industrial warehouse and a high-intensity AI factory blurs.

Residents are increasingly reporting a decline in livability. In Melbourne, locals have cited constant construction noise and the insistent background hum of servers. More concerning is the reliance on backup power; at the M3 site, diesel generators are reportedly expanding from 40 to 100 at completion, raising alarms about local air quality.

The “Cluster Effect” and Urban Pressure

In areas like Lane Cove, Sydney, the issue isn’t just one building, but the “cluster effect.” In some local industrial zones, datacentres already take up 40% of the land. When new proposals like “Project Mars”—a 90MW facility—emerge, they often sit adjacent to critical community assets.

Local residents, such as Daniel Bolger, have pointed out the irony of placing these industrial giants next to the lungs of Lane Cove, referring to Blackman Park, a sporting hub used by 50% of the suburb each weekend.

Beyond Diesel: The Quest for Sustainable Compute

The future of AI infrastructure depends on solving the “energy paradox”: AI needs massive power, but the community will no longer tolerate the pollution that comes with traditional backup generators.

We are likely to witness a trend toward “Green Compute” zones. This includes a shift away from diesel toward hydrogen fuel cells or on-site renewable microgrids. Without this transition, the “environmental cost” of AI will continue to trigger fierce community opposition.

In Western Australia, the tension is centered on ecological preservation. Proposals for a 120MW facility in Hazelmere have met resistance from the Bibbul Ngarma Aboriginal Association, with chair Walter McGuire stating,

“Giant datacentres belong in industrial areas, not on the banks of our rivers and wetlands.” Walter McGuire, chair of the Bibbul Ngarma Aboriginal Association

Sovereign AI vs. Local Livability

The push for these facilities is often framed as a matter of national security and economic independence. The industry argues that Australia must build its own infrastructure to avoid becoming a mere importer of foreign technology.

Belinda Dennett, CEO of Data Centres Australia, has emphasized that without domestic AI infrastructure, the country risks relying on technology that has no Australian culture, values or laws built into that.

The challenge for the next decade will be balancing this “Sovereign AI” ambition with the rights of the citizens living in the shadow of the servers. The trend is moving toward a requirement for more transparent, merit-based assessments that prioritize ecological and psychological health over mere speed of deployment.

For more on the intersection of tech and environment, explore our guide on the carbon footprint of generative AI or read about the evolution of smart city zoning.

Frequently Asked Questions

What is a hyperscale AI factory?
It is a massive datacentre specifically designed to handle the immense computational power required for training and running large-scale AI models, utilizing high-density GPU clusters.

Why are residents opposing AI datacentres?
Common concerns include noise pollution (the “hum”), air pollution from diesel backup generators, visual impact on the skyline, and the encroachment on parks and wetlands.

What is “Sovereign AI”?
Sovereign AI refers to a nation’s ability to produce its own artificial intelligence capabilities using its own infrastructure, data, and workforce, ensuring the technology aligns with local laws and cultural values.

How much power do these facilities employ?
Power requirements vary, but recent Australian proposals range from 90MW to 225MW, placing significant strain on local electrical grids.