On 30 March 2026, MPI and Murdoch University submitted an objection to DA-6/2026, a proposed three-storey, 23.5-metre data centre at Lots 14 and 15 Stirling Crescent, Hazelmere, in the City of Swan. As far as we know, it is the first time a mining policy organisation formally objects to a data centre on the basis of mining governance principles.
We were alerted to the proposal by Bibbul Ngarma Aboriginal Association, a Noongar Traditional Owner-led organisation working on the BoorYul-Bah-Bilya program – an integrated management plan for the Mandoon Bilya (Helena River) catchment. BoorYul-Bah-Bilya is the first program of its kind to consider environmental, social, cultural, and economic values in a single integrated river plan, and it is timed to coincide with the 2029 Perth Bicentenary. Bibbul Ngarma’s chairperson, Walter McGuire, a Noongar Traditional Owner, has described the work as reclaiming a responsibility that has been denied since colonisation: “It’s our role and responsibility as the First People of this land. Since colonisation we’ve been denied our responsibility and role so we’re bringing it back through this for people to understand this is how things have been done for millennia.”
The proposed data centre sits within the Helena River catchment – the same landscape Bibbul Ngarma is working to restore – and immediately adjacent to Trillion Trees Australia, a community nursery that has planted over 15 million native trees from Hazelmere since 1979. Bibbul Ngarma’s findings about cumulative pressures on the catchment, combined with what we know about data centres’ resource demands, prompted our submission.
What does a mining watchdog have to do with a data centre?
More than you might think.
MPI has spent thirty years working on the governance of resource-intensive industries. We assist communities affected by mining, and we evaluate projects against the Mining Policy Principles – a set of non-negotiable standards developed collaboratively by over a dozen Australian civil society organisations, including the Australian Conservation Foundation, The Wilderness Society, Transparency International Australia, and the Conservation Council of WA.
The more we examined the data centre expansion, the more clearly the structural parallels with mining emerged. Data centres are industrial-scale operations that extract shared public resources – energy, water, land, and community amenity – to generate wealth that flows overwhelmingly to distant shareholders, while concentrating costs and impacts on local communities. The community provides the resource. The profit flows elsewhere. The community has no say in what the resource is used for.
This is the structural logic of mining. And the principles the mining reform movement has fought for – cumulative impact assessment, regional planning, community self-determination, transparency, no-go zones – apply directly.
The intelligence curse
Our submission foregrounds what researchers Luke Drago and Rudolf Laine call the “intelligence curse” – the AI-era equivalent of the resource curse that has plagued mining-dependent economies for decades.
The resource curse is well-documented in the mining literature. In Western Australia specifically, Brueckner et al.’s Resource Curse or Cure? (2014) examined how the state’s mineral boom concentrated wealth while leaving communities bearing inflated costs of living, environmental degradation, and unfulfilled promises of lasting benefit. Roche and Mudd’s contribution to that volume – ‘An Overview of Mining and the Environment in Western Australia’ – documented the cumulative environmental toll: declining ore grades demanding more energy and water per tonne of output, expanding waste footprints, and a regulatory architecture that consistently lagged behind the pace of extraction. Charles Roche is now MPI’s Executive Director, and the patterns he and Mudd identified in mining a decade ago are remarkably legible in the data centre expansion today.
The intelligence curse extends this logic: once powerful actors can generate wealth from AI rather than human labour, they lose the structural incentive to invest in people. Education, employment, community infrastructure – all become costs without returns. Tristan Harris, co-founder of the Center for Humane Technology, recently described this on US national television as a trajectory toward “an anti-human future.”
The Hazelmere proposal makes this concrete. The facility includes just 41 parking bays – a reliable indicator of minimal permanent employment. Meanwhile, the AI systems these data centres power are actively displacing workers. Block Inc. has laid off 4,000 employees citing AI automation. Atlassian – an Australian company – cut 1,600. The data centres being approved today are the infrastructure that will power tomorrow’s redundancies in the very communities providing the resources.
Water: the sharpest point
Perth is one of the world’s most water-stressed capitals. Dam inflows have collapsed approximately 80% since the 1970s. The Integrated Water Supply Scheme sourced 337 billion litres in 2024-25 – and is already reaching its limits, according to Water Corporation. By 2050, existing infrastructure will supply less than 200 gigalitres per year, while demand is forecast to reach 438 gigalitres.
Into this context, data centres are arriving with significant and poorly disclosed water demands. Existing community analysis of the Hazelmere proposal drew on European water consumption figures. We flagged that these almost certainly understate consumption in Perth’s climate. Data centre cooling depends heavily on ambient temperature. “Free cooling” (a misnomer if there ever was one - see further) using outside air only works in cooler regions – Perth, with summers regularly exceeding 35°C, does not qualify. Even closed-loop systems consume more energy to reject heat in hot conditions, and that energy carries its own embedded water cost. Industry analysis of Australian data centre cooling notes that global providers often rely on standardised reference designs from cooler regions, but that Australia’s climate demands fundamentally different assumptions.
In Melbourne’s west, 19 proposed data centres are seeking 20 gigalitres of water annually – enough to supply 330,000 people. WA has no regulatory framework to prevent a similar trajectory, and no mandatory reporting to even measure it, making it a target for developers.
What makes this worse is that the absence of regulation is not an oversight – it is a selling point. A peer-reviewed study published in Water in January 2026 found that the only WA government document addressing data centres was a 2022 marketing prospectus produced by the Department of Jobs, Tourism, Science, and Innovation – titled WA: The southern hemisphere’s global hub for data centre operations – which actively promotes the state’s “water availability” to attract data centre investment. The researchers found no corresponding regulatory or environmental framework. WA is marketing its water to an industry it has no mechanism to regulate. For anyone familiar with the history of mining in this state, the pattern is unmistakable: attract the investment first, worry about the governance later – if at all.
The physics: both mining and AI are entropy machines
The parallel between mining and data centres runs deeper than governance. It is physical.
Both are high-entropy activities. Mining takes concentrated ore and disperses it – turning ordered geological deposits into diffuse waste, tailings, and emissions. Data centres take concentrated electricity and disperse it as heat. In both cases, the Second Law of Thermodynamics is doing the driving: useful energy is being irreversibly degraded into waste. This is not unique to mining or data centres – it is how all energy conversion works. The question is not whether energy will be dissipated, but whether any given dissipation is worthwhile, and whose interests it serves. A hospital dissipates energy. So do weapons systems. The physics is the same; the politics could not be more different. Communities have every right to ask: what is this energy being spent on, and who benefits?
This matters because the industry language around data centre “cooling” is fundamentally misleading. Cooling does not make the heat disappear. It cannot. Thermodynamics forbids it. Every watt of electricity consumed by a server is converted into heat that must go somewhere – into the air, into water, into the ground. The choice is not whether to heat the local environment, but how: via evaporated water (wet cooling), via hot air (dry cooling), or via some combination. “Zero water for primary cooling” – a claim made by CDC for its Maddington campus – simply means the heat is rejected into the air instead of into water vapour. The heat is the same. The local thermal load is the same. The entropy increase is the same.
In a city already experiencing urban heat island intensification under climate change, this is not a technicality. It is a material environmental impact that “sustainability” marketing systematically obscures. A 14,883m² industrial facility processing data at scale will concentrate significant waste heat in a suburban-industrial area adjacent to a river catchment and a community nursery. No amount of engineering changes the thermodynamics – it only changes where the heat lands.
This is a thread we will return to in future editions of the Mining Monitor. The connection between mining and AI is not merely metaphorical. Both are dissipative processes – they consume concentrated energy and materials to produce economic ‘value’, and in doing so they irreversibly degrade the environment’s capacity to sustain life. Understanding this shared physics is, we believe, essential to understanding why both industries require the same governance scrutiny, and why communities are right to insist on it.
Mining principles applied
Our submission maps the Mining Policy Principles (2025) directly onto the Development Application:
Cumulative impacts. This proposal does not exist in a vacuum. It arrives alongside the CDC Maddington 200MW campus – the largest data centre planned for Western Australia – and a pipeline of further proposals across Perth’s eastern corridor. The Council is approving facilities one at a time, with no assessment of cumulative impact on the electricity grid, water supply, or community character.
Regional planning. WA has no data centre policy, no strategic framework for digital infrastructure, and no regional assessment process. Approving data centres without that framework is the planning equivalent of approving mines without understanding the geology of the region.
Community self-determination. The Mining Policy Principles distinguish between consultation and consent. The community of Hazelmere was given a comment period on a proposal already designed and lodged. They were not asked whether a data centre was appropriate for this site. As with mining, the community bearing the costs has no visibility into what the facility computes – consumer queries, AI training, surveillance systems, or military applications.
Transparency. All supporting documents were commissioned by the applicant. No mandatory Water Usage Effectiveness (WUE) reporting exists in WA. There is no independent mechanism to measure or limit actual water consumption over the facility’s life.
No-go zones. The site sits adjacent to Trillion Trees and within the Helena River catchment that Bibbul Ngarma is working to protect. A 23.5-metre industrial facility here raises direct questions about ecological red lines.
People in the US are already pushing back
The political argument may be the most immediately relevant for local government.
In the United States, where the data centre industry is most mature, at least 25 projects were cancelled in 2025 alone due to local opposition – four times the number in 2024. An estimated US$64 billion in projects have been blocked or delayed. There are now 142 organised advocacy groups in the 28 states where there are >50MW data centres projects proposed. More than 230 organisations have called for a national moratorium on new data centre construction.
This resistance is bipartisan. In Texas, the state Senate backed tighter regulation. In Ohio, local governments imposed moratoriums. Only 44% of Americans would welcome a data centre in their neighbourhood – making data centres less popular than gas plants, wind farms, or even nuclear facilities.
Communities that approved data centres early are now seeking moratoriums and regulatory tightening. The City of Swan has the advantage of learning from the US experience rather than repeating it.
What we’re asking for
We have urged the Council to refuse DA-6/2026, or at minimum defer consideration. Any future assessment must centre the self-determination of the Noongar Traditional Owners whose Country this is, and must address the following:
1. A state-level strategic framework for data centre development in Western Australia. The City of Swan should not be making decisions of this magnitude in isolation. Data centres are critical infrastructure with significant energy, water, and land implications that cross local government boundaries. Western Australia needs a state-level policy – equivalent to what NSW is now developing through its Legislative Council inquiry – that sets cumulative impact thresholds, mandates environmental referral for facilities above a given capacity, and establishes clear criteria for site selection, water sourcing, and community benefit. In the absence of such a framework, local councils are being asked to approve industrial developments whose consequences are regional and state-wide, using planning instruments designed for sheds and warehouses.
2. Free, Prior and Informed Consent. The DA documents include an Aboriginal Heritage Due Diligence Assessment and an Archaeological Survey, but compliance with heritage clearance procedures is not the same as genuine consent. Bibbul Ngarma Aboriginal Association is actively working to restore and protect the Mandoon Bilya (Helena River) catchment through BoorYul-Bah-Bilya – a Noongar-led integrated management plan. Industrial development within that catchment should not proceed without the meaningful agreement of the Traditional Owners leading that work. The Mining Policy Principles are unambiguous on this point: FPIC, including the right to say no, must apply to all resource-intensive development on Country – not only to mining.
3. A comprehensive data centre policy for the City of Swan, including cumulative impact assessment, site selection criteria, and genuine community engagement – not just comment periods on individual applications already designed and lodged.
4. Mandatory, independently audited Water Usage Effectiveness reporting calibrated to Perth’s climate – not imported from European or US temperate baselines. WA currently has no mechanism to measure, verify, or limit actual water consumption by data centres. As we have noted, the state government is actively marketing water availability to attract the industry while maintaining no regulatory framework to manage it.
5. Cumulative impact assessment of the combined effect of this proposal, the CDC Maddington 200MW campus, and the broader pipeline of data centre developments on Perth’s energy grid, water supply, urban heat, and community amenity.
This submission was an initial response, prepared at short notice. We intend to develop these arguments further. But we believe the core claim is already clear: data centres are a new form of resource extraction, and they deserve the same rigorous, independent, community-centred scrutiny that the mining reform movement has spent decades fighting to establish.
We are grateful to Francesca Flynn - Bibbul Ngarma Aboriginal Association’s executive director - for her diligent submission to the City of Swan, which alerted us to this proposal. We are also grateful for the work the BNAA is doing on Mandoon Bilya – work that embodies the kind of community-led, values-based stewardship that should be shaping land-use decisions in Hazelmere, not being displaced by them.
May we choose our entropy wisely.