Western Australia is being asked to make room for a new class of infrastructure: buildings that may look like warehouses, employ relatively few people once operating, and yet consume resources at a scale more familiar to heavy industry.
Data centres are now essential to cloud computing, artificial intelligence, cyber resilience, sovereign data storage and the digitisation of mining, energy, defence and research. Western Australia has good reasons to want them. The State has promoted WA as a data-centre destination, pointing to political stability, international submarine-cable links, renewable-energy potential and proximity to Asia.
Recent developments have sharpened the issue. In August 2026, the Australian Energy Market Commission advised that large data centres should bring new clean, firmed energy, connect efficiently and pay their way. NSW also linked streamlined assessment to energy, water and community infrastructure commitments. For WA developers, infrastructure strategy is becoming part of the approvals pathway.
The opportunity is not confined to Perth. Regional WA may offer large sites, renewable resources and proximity to mining and energy operations. A proposed AI data centre south of Broome has brought the Kimberley into the conversation. But regional projects face different tests, including power, fibre, water, workforce, emergency response, heritage and Country.
But economic importance does not make a use impact-free, and an industrial zoning label does not make every industrial site suitable. The central planning mistake would be to assess a data centre mainly as a large building. Its real footprint extends well beyond the lot boundary - into the electricity network, water system, fibre network, emissions inventory and the supply of strategically located land.
Not a warehouse with servers
Traditional development assessment is strongest where impacts are visible and local: traffic, parking, setbacks, height, landscaping, streetscape and overlooking. Data centres invert that pattern. Their traffic generation can be modest and their façades quiet, while their least visible inputs (firm electricity, cooling, network redundancy and backup generation) can be the most consequential.
That mismatch begins with classification. Data centres are not uniformly defined in Western Australian local planning schemes, which can force them into an existing category such as industry, warehouse, office or an unlisted use. None is a comfortable fit.
In July 2026, the City of Vincent began considering whether data centres should be separately defined, where they should be permissible, and whether local guidance is needed for noise, heat, servicing, built form and sensitive interfaces. That is not a niche planning question. Classification determines where a proposal can locate, how it is advertised, which objectives guide discretion and what evidence is expected at application stage.
A State-level definition and assessment framework would provide greater consistency, but it should distinguish between very different facilities.
A small facility embedded in a commercial building is not equivalent to a 100-megawatt hyperscale or AI campus. A sensible framework would scale requirements to power demand, water demand, generator capacity, site area and proximity to sensitive uses, not simply floor area or development value.
Electricity is the first site-selection and land-use question
For a large data centre, access to electricity is often one of the main factors that determines whether a site is suitable. This is particularly important in Western Australia, where a proposal may rely on the isolated south-west system or require dedicated regional generation, firming and network infrastructure.
Planning authorities do not need to become electricity experts. But they do need confidence that the project has been discussed with the relevant electricity providers and that its likely power needs are understood. Without that early work, a proposal may receive planning approval only to face long delays, major infrastructure costs or changes to its design later.
There is also a broader public question about how new infrastructure is funded and whether large new users could add pressure to the system. These matters are mainly for government, electricity agencies and the industry to resolve. Recent WA Budget investment in clean energy and network expansion is not data centre-specific funding. Developers still need to establish which enabling works are committed, who pays and how delivery will be staged. The role of planning should be to bring those parties together early, make the key assumptions visible and ensure that development occurs in step with available infrastructure.
Water use is a design choice
Water use is another issue that can be difficult for the public to understand. Not every data centre uses the same amount of water. The demand depends on the facility’s design and the way it is cooled. That level of technical detail should be assessed by water providers and suitably qualified specialists.
From a planning perspective, the higher-level questions are more straightforward. How much water is likely to be needed? Is drinking water required, or could recycled water be used? Is that supply available at the site? Could the project place pressure on other users, particularly during hot and dry periods?
These questions matter across WA, although the answer will differ between Perth and regional sites. In water-constrained locations, a project should demonstrate that its cooling model will not place unreasonable pressure on other users. The expansion of the Kwinana Water Reclamation Plant shows how recycled water can support new industry without relying entirely on drinking water. Similar thinking could help guide data centres towards areas where water can be supplied responsibly.
The impacts that do cross the fence
The recent Hazelmere proposal is a warning against describing data centres as ‘quiet’ infrastructure without qualification. The proposed 120-megawatt facility was withdrawn in May 2026 after strong community opposition and a recommendation for refusal. Reported issues included emergency-generator noise near a school and homes, together with the site’s relationship to a conservation wetland, the river and Aboriginal heritage places.
Backup generators may operate infrequently, but they are tested routinely and exist to run during abnormal conditions. Noise assessment should therefore address testing, maintenance and credible outage scenarios; cumulative operation of generator banks; time of day; tonal characteristics; exhaust and air quality; and nearby sensitive receptors. Battery systems introduce a different risk profile, including fire and emergency response. Mechanical plant can also create persistent tonal noise, and heat rejection may affect the immediate microclimate.
Security-led design creates another planning tension. Blank walls, setbacks, fencing, limited entrances and low staffing may be operationally rational but can weaken street activation and passive surveillance. Those characteristics are generally more compatible with well-buffered industrial precincts than mixed-use centres, transit-oriented areas or scarce inner-urban commercial land expected to support employment and active frontages.
Land efficiency also deserves more attention. Operational job numbers can be low relative to site area, while electricity reservations and infrastructure corridors may sterilise capacity that could support other industries. This is not an argument to reject data centres because they employ fewer people than a conventional business park. Their economic contribution is different and partly indirect. It is an argument to assess opportunity cost explicitly, especially on constrained strategic industrial land and sites earmarked for higher-density urban outcomes.
From data centre development applications to infrastructure planning
WA does not need a policy designed to stop data centres. It needs one capable of choosing the right facilities, in the right places, with costs and impacts disclosed early. Five reforms would materially improve the system:
- Define and tier the use. Introduce a consistent data-centre definition, with thresholds based on electrical load, water demand and backup generation so assessment is proportionate to scale.
- Identify infrastructure-ready precincts and regional locations. Map transmission capacity, substations, diverse fibre routes, recycled-water availability, wastewater capacity, hazard exposure, buffers and strategic land constraints. Zoning alone is not a servicing strategy.
- Require a resource and resilience statement. Major proposals should quantify staged electricity and water demand, cooling technology, renewable-energy and storage arrangements, grid-connection status, generator operation, emissions, waste heat and end-of-life equipment management.
- Assess cumulative effects. Multiple facilities drawing on the same substation, water source, fibre corridor or industrial precinct should not be assessed as unrelated projects. Precinct-scale thresholds should trigger coordinated infrastructure review.
- Link approval to delivery. Conditions and staging should prevent occupation or energisation beyond verified network and water capacity, while allowing early works where connection pathways are credible.
The heavy industry of the digital age
Western Australia’s opportunity is real. Local and regional data infrastructure can strengthen sovereign capability, business continuity and the competitiveness of WA’s resource, defence, research and technology sectors. A steady, flexible load may also support investment in new generation and storage if the infrastructure is genuinely additional and well-coordinated.
But the public case cannot rest on the idea that digital infrastructure is inherently clean because its output is intangible. Servers turn electricity into heat. Cooling can transfer the burden between power and water. Redundancy brings generators, batteries and duplicated networks. Large campuses consume land and can reshape the investment priorities of public infrastructure providers.
Data centres should therefore be planned as the heavy industry of the digital age: essential, capital-intensive and potentially transformative, but resource-hungry and highly sensitive to location. The facilities may continue to hide in plain sight. Their infrastructure demands should not.
For developers considering metropolitan, strategic industrial or regional sites, early planning due diligence is critical. TBB Planning can help test land-use permissibility, site constraints and infrastructure assumptions, and shape the development application and planning approvals pathway before key decisions are locked in.
Learn more about our Industrial specialism here
Data centres should therefore be planned as the heavy industry of the digital age
8 min read
