'He is growing, neurologically, pulmonarily, physically, in the shadow of a facility whose cumulative environmental impact has never been assessed,' says Sean Brown, a West Footscray resident who walks his toddler past NextDC’s M3 campus. The proposed expansion near Melbourne has become a focal point for rising community pushback. Residents complain of constant construction noise, a low-level hum and diesel exhaust as the project edges closer to homes, parks and schools. NextDC describes the site as "Australia’s largest hyperscale AI factory" while local councils and a Victorian planning spokesperson say the proposal is under formal review.

NextDC’s M3 campus in West Footscray sits under 10km from Melbourne’s CBD. The company says it plans to expand the site to about 10 hectares and to draw roughly 225MW of power if the Victorian government grants fast-track approval and if the build proceeds to completion by the end of 2027.

Neighbours say the build is already too close

Nearby residents describe an everyday rhythm that has shifted. "He is growing, neurologically, pulmonarily, physically, in the shadow of a facility whose cumulative environmental impact … has never been assessed," said Sean Brown, a West Footscray resident who walks his toddler past the site. He pointed to noise from construction, a constant background hum and the smell of diesel from backup generators.

On the site, diesel generators are reported to be expanding from about 40 today to roughly 100 once the campus is complete. Locals say that increase matters because the generators run as on‑site backup for servers that operate 24 hours a day.

What developers say

NextDC’s chief executive, Craig Scroggie, posted a video showing rapid construction and described the project as "Australia’s largest hyperscale AI factory purpose‑built for the new AI era of accelerated computing." The company frames the campus as a way to deliver sovereign digital capacity, local jobs and infrastructure for cloud and AI services.

That pitch mirrors industry arguments heard elsewhere: large datacentres support national competitiveness in cloud and AI services and create business‑grade capacity that global customers want. Developers point to regulated approvals and technical standards as proof the projects meet environmental and planning tests.

Where planning power lies

Local government has pushed back. Maribyrnong council has objected to the scale and proximity of the M3 expansion to residential areas. Still, final approvals sit with state planning authorities.

A Victorian government spokesperson for Planning Minister Sonia Kilkenny said the M3 proposal was under consideration.

That division of responsibility is central to the conflict. Councils field direct complaints from neighbours about noise and air quality. State planners weigh broader economic arguments, land‑use rules and grid capacity. Residents say the original zoning decisions didn't anticipate the current scale of hyperscale facilities.

Mastheads

Wider patterns: why this is happening now

New international demand for AI training and cloud services has altered developers’ calculus. Large, energy‑intensive facilities cluster where land, grid connections and favourable planning combine. The Harvard Gazette noted there are already more than 4,000 datacentres in operation in the United States, mostly in Virginia, Texas and California, with some 3,000 more planned or under construction.

Those numbers reflect a shift in what counts as a standard datacentre. The Guardian and academic sources describe modern hyperscale sites as far larger than older definitions. A common industry benchmark defines hyperscale as more than 10,000 square feet and more than 5,000 servers, but current builds often exceed that minimum by a wide margin.

Residents and researchers list a similar set of worries. They include noise from cooling infrastructure and backup power, increased local air pollution from diesel generators, heavy power demand on local grids, and large water use for cooling at some sites. The Harvard interview with Ben Green, assistant professor at the University of Michigan School of Information and Public Policy, said public concern is legitimate and often focused on rising electricity rates, water consumption, and tax incentives for developers.

Green also argued that datacentres don't reliably bring large numbers of local jobs. He said communities have pushed back effectively in some places, blocking projects or prompting municipalities to adopt moratoria that pause further developments.

Developers counter that datacentres create property tax revenue, some local construction employment and the digital backbone needed for businesses that rely on cloud services. Critics say those benefits are uneven and often don't match the scale of environmental and community costs.

At the centre of the debate is infrastructure. Datacentres need steady, large‑scale power. They also need reliable backups, and in practice that often means diesel or other generators on standby. Where grid upgrades lag, operators rely on on‑site backup fuel to avoid service interruptions.

That dynamic shapes planning decisions. If state planners approve large campuses near built‑up suburbs, the direct impacts, noise, additional truck movements, visible infrastructure and backup emissions, fall on neighbouring communities. Where planners steer datacentres to industrial zones with stronger buffering, neighbours report fewer disruptions.

Civic organising has become a visible part of the trend. Community groups have won moratoria in some US jurisdictions and have blocked individual projects. In Australia, residents around M3 and other proposed sites have lobbied councils and state planners, raised legal objections and drawn media attention to local effects.

Ben Green said public mobilisation has mattered. "It’s been impressive just how well communities have organised around this," he said. "Many datacentre projects have been blocked by local resistance, and many municipalities have passed moratoria that pause datacentre development."

The debate makes a clear tradeoff visible. Building large datacentres in or near cities delivers capacity for AI and cloud services close to users. That can reduce latency and attract investment. But it also concentrates heavy power use and industrial infrastructure near homes and schools.

That contrast is what fuels the current backlash. Residents living next to sites like M3 see daily effects. Councils and planners must weigh those community costs against state and national economic arguments.

Related Articles

A Victorian planning spokesperson said the proposal remains under formal review by state planners, who will decide whether it can proceed. NextDC said: "We’re building Australia’s largest hyperscale AI factory purpose-built for the new AI era of accelerated computing."

This article was created with AI assistance.