Every swipe, tap and click runs through a data centre. We check the weather, scan the news, load a map — and somewhere, a building full of servers does the work. The question planners have a role in is where the buildings should go and who benefits.

Data centres need a land use response

Data centres are positioned as enabling economic infrastructure that will bring growth, investment and productivity. The recent debate has focused on the AI they run, or on the energy and water it takes to drive them. Another important focus needs to be: where should they go, and who benefits?

These are classic planning questions, and data centres are the most significant land use and infrastructure challenge we have faced in decades. Location, economic opportunity, cumulative impacts, community benefit, social licence — this is a planner's day-to-day work. Investment at this scale offers an opportunity to deliver long-term benefits to the places and communities where it occurs. These benefits will only transpire with strategic planning and a focus on broader economic opportunities.

The scale demands action

Australia's data centre pipeline is estimated to exceed A$150 billion by 2030 [1], around 5.6% of national GDP. In NSW, the Investment Delivery Authority has moved 15 projects worth A$51.9 billion into its pipeline and screened out a further A$40.7 billion as speculative. Sector investment has grown at roughly 65% a year and now makes up about 12% of non-residential building activity in NSW [2].

Two forces are driving this increased investment. The first is the steady growth of everyday digital activity and the second, more recently, is AI, which is changing the scale and intensity of what gets built.

History gives us reason for care. Canal mania, the 19th-century railway boom, and the late-1990s fibre buildout all followed a similar pattern: heavy up-front capital, then challenges when returns don’t match expectations. In the USA, railway investment peaked at around 6% of US GDP, fibre at about 1% [3]. McKinsey indicates global investment in AI and data centres could reach close to US$7 trillion by 2030 [4].

These earlier large economic investments left assets (canals and railways) that still stand today and shaped our cities and regions over decades. This contrasts with the chips inside a data centre, which can become obsolete in 2 to 5 years [5]. What we are seeing with data centres is a more volatile, more capital-intensive pattern that is unfolding over a handful of years rather than decades. The risk is that the gains from this investment concentrate elsewhere while communities bear the costs of data centres. Governance and investment settings, including guidance on where data centres should go, will determine where the benefits flow.

Location matters

Three types of facilities are being built, and they cannot be treated the same. Edge data centres are small, decentralised and need low latency, so they sit close to users to reduce delays. Hyperscale and enterprise facilities also stay near users and run trained models. AI campuses train models at high compute density, where inference latency is irrelevant and large sites with dedicated power are preferred.

All of the three types of data centres seek to be close to four scarce things: energy transmission lines, fibre networks, subsea cable routes, and water for cooling. The largest-scale AI campuses also seek out very large sites.

In Greater Sydney, the subsea connection to Singapore makes for an attractive location. Within Sydney, investment in data centres is landing in industrial precincts — Western Sydney, Macquarie Park, and the Lower North Shore — with a focus on those near energy transmission lines. Data Centres above 250 MW need nearby higher-voltage infrastructure to support operations. Location attributes are also evident in other investment decisions nationally, with Tasmania's cool climate and hydroelectric power sought by AI training campuses, and the ACT remaining a niche sovereign data enclave where end users require proximity and high security.

Industrial lands were planned with the needs of freight and logistics operators and manufacturing operations in mind. Their location and supporting infrastructure have been designed specifically for these uses. Similarly, industrial lands for urban services are close to the populations and businesses that use them. Western Sydney Regional Organisation of Councils, Northern Sydney Regional Organisation of Councils, and the National Growth Areas Alliance, in their submissions to the NSW Parliamentary Inquiry into Data Centres, all noted their support for data centre investment. They also made clear that these investments compete directly with planned economic activities on industrial lands. In particular, data centres tend to seek larger sites and are out-competing freight and logistics users. Recent data centre proposals in Northern Sydney estimate that they will generate 25 to 30 operational jobs, while industrial precincts in the region generally achieve double these figures. Without planning for this overall increase in demand for industrial lands, the job capacity of existing industrial land is inadvertently being reduced. So far, they are less embedded or connected to the local economy, with weaker supply chain linkages and agglomeration potential. There is very little understanding of these cumulative impacts.

Recent new policy responses

This year has brought real policy movement, with more to come next year. National Cabinet has confirmed a nationally consistent framework setting minimum requirements for large data centres across energy, water and land use by early 2027 [7]. The Prime Minister has pointed to our comparative advantage — "enormous space, which is the envy of the world" [8] — alongside the need to keep energy bills down and water security intact. The point being missed is that we need more than space; we need connections to the required infrastructure. Data centres, depending on their scale, seek out large sites with low latency and high-capacity energy transmission. These attributes require integrated planning and investment.

The NSW Guidelines [9] are silent on location, beyond noting that NSW is an appealing place to invest. The Sydney Plan commits to “Identify potential new industrial land, including identifying suitable industrial lands for data centres" [10].

The recently released Victorian “The Sustainable Data Centre Action Plan” [11] provides locational controls, including a buffer between residential buildings and proposed data centres (150M), and a prohibition in rural areas unless they can be serviced by data centre infrastructure. Statewide planning policy to guide location, design, scale and infrastructure needs will be developed in 2027.

Case study: Northumberland Council, UK

Northumberland, a rural county of 320,000 people in north-east England, is using a landmark hyperscale data centre investment to reposition its economy, previously built on coal, shipbuilding and heavy industry.

Northumberland’s former Cambois power station is a strategic site with existing high-voltage power infrastructure, a deep-water port, fibre connectivity, and proximity to industrial supply chains. Energy availability, not connectivity, was the decisive locational factor. Blyth was one of the first designated AI Growth Zones in the UK [12] and investment in a hyperscale data centre campus was secured. Northumberland Council also held a legal charge over the land after the previous owner entered administration. They used that position to convert a locational advantage into public value through an agreement for a Growth Fund of up to £110 million over an 8–10 year build-out, tied to development milestones, with around £18 million already received.

The Council's objective was to avoid the data centre becoming "a black box that sits there". The Growth Fund prioritises enabling infrastructure for employment lands, and a recoverable fund for local business growth. This focus on regional economic development was a priority.

In addition, social value commitments are embedded in construction contracts and supply chain requirements, including local recruitment, skills and training, and community investment. Project-driven grid upgrades removed one of the county's main barriers to industrial development. The skills needed for data centre construction and operations overlap with offshore renewables and advanced manufacturing, so workforce investment carries across sectors and benefits the region, rather than sitting in a silo.

The Council's Economic Strategy to 2035 includes “Digital and AI” as one of 5 priority sectors. Data centres are part of the overall value chain, which also considers supporting infrastructure and sector use cases and applications. The strategy builds on Northumberland’s strategic location, connectivity and growing digital infrastructure to gain further momentum in software development, analytics, AI applications, data centre hosting and cybersecurity. Council is also developing an employment lands strategy to align supply with future needs. Council will assess any future data centre proposals on long-term jobs, local investment and social value. This ensures hyperscale anchors are distinguished from speculative builds and that powered employment land is not consumed without wider economic return. Given the nature of the investment and technology cycles, Council is also considering make-good requirements if uses become redundant.

The case study demonstrates that the economic benefit is beyond the facility itself. Local leadership is facilitating and aligning longer-term integrated opportunities [13].