Discussion of artificial intelligence concentrates on models, governance and liability. How much capacity Australia actually builds will be settled at ground level: whether a site can be lawfully zoned, connected, cooled and lit.

A different kind of constraint

Australia's data centre sector is growing faster than the infrastructure that enables it. Announced investment pipelines are measured in tens of billions of dollars. The foundational inputs every facility requires (land, power, water and fibre) are each governed by legal and regulatory frameworks that were not designed with hyperscale development in mind.

This means the limiting factor on AI capacity in Australia is not technological and, for the most part, not financial. It is legal and regulatory: whether a site can be characterised as a permissible use, whether a connection offer can be obtained on viable terms and within a viable timeframe, whether water can be secured without displacing community support, and whether diverse fibre can be delivered to the campus boundary.

For in-house counsel and senior executives, understanding where the friction sits within each of these constraints is a prerequisite for credible project planning and realistic board-level commitments.

The technology is new. The frameworks that determine where it can lawfully be built are not, and they were not written for it.

Land: the supply of legally available sites

The permissibility problem

The starting point for any site assessment appears simple: is the proposed use permitted on this land? For data centres, the answer is frequently unclear. Planning instruments across Australian jurisdictions were drafted before hyperscale data centres existed as an asset class, and the classification of the use, whether as an industrial facility, a utility installation, a technology precinct use, or something else, varies between instruments and is sometimes genuinely ambiguous.

In New South Wales, the Standard Instrument (Local Environmental Plans) Order 2006 does not include ‘data centre’ as a defined land use. Characterisation must instead be determined from the particular facility’s purpose, physical characteristics and operations, read together with the definitions and land use table in the applicable LEP. It may differ between projects, and should not be assumed from the presence of telecommunications equipment alone. Victoria’s planning schemes, and the industrial and special purpose zones in Queensland and Western Australia, present analogous questions. Developers operating across jurisdictions need jurisdiction-specific advice on characterisation before committing to a site.

Zoning constraints and rezoning

Even where a use is permissible, zoning controls frequently limit building height, floor space, site coverage, setbacks and noise emissions in ways inconsistent with large-scale data centre operations. Projects must navigate industrial or employment zone provisions designed for warehousing and light manufacturing, not hyperscale facilities with continuous high-load operations and substantial mechanical plant.

Where rezoning is required, developers face a multi-stage statutory process involving council, state agencies and, in some cases, the Minister for Planning. Rezoning can take several years, although the timeframe varies substantially according to the statutory pathway, site complexity, agency consultation and whether the proposal attracts state-level intervention. Developers who acquire sites carrying rezoning risk without a clear legal pathway and a realistic timeline frequently find themselves holding land they cannot develop within their investment horizon.

Even where permissibility is not in issue, approval timelines require active management, including through merit review appeals in jurisdictions where those are available.

Title-level constraints

Title constraints can render an otherwise appropriately zoned site legally problematic. Contamination is the most common issue on industrial land, where prior use may carry remediation obligations under the Contaminated Land Management Act 1997 (NSW) or its equivalents in other states. Acquiring contaminated land can expose a developer to substantial investigation, management and remediation costs, together with reporting and compliance obligations. The extent of that exposure depends on the applicable legislation, the significance and source of the contamination, the proposed use and the contractual allocation of risk.

Heritage overlays, biodiversity certifications, utility easements and restrictive covenants require careful pre-acquisition due diligence. Data centre footprints are large and inflexible. Unlike a commercial office development, a hyperscale facility cannot readily accommodate an unexpected easement corridor or a heritage-protected building envelope, which makes title review more consequential than in most other asset classes.

Site assembly

In established metropolitan areas, sites large enough for hyperscale development are often in fragmented ownership, requiring assembly from multiple vendors and introducing coordination and holdout risk. The common approach is conditional contracts with each vendor, contingent on all other parcels being secured, with long sunset dates and termination rights that address planning risk as well as physical site risk.

Power: connection, cost and cost allocation

Grid connection as a legal and commercial constraint

For most large data centre projects in the current market, the electricity grid connection process is the single most consequential legal and commercial constraint on delivery. The National Electricity Market rules governing connection (administered by AEMO through the frameworks applying to Transmission Network Service Providers and Distribution Network Service Providers) were designed for a market in which large new loads appeared infrequently and incrementally. They are poorly suited to a pipeline in which dozens of major loads, each seeking 100 MW or more, present for connection simultaneously across a limited number of viable corridors.

Connection queues are congested. In some regions the queue for new major connections extends for several years. Applicants receive offers that may not be commercially viable, and negotiating a revised offer, or lodging a dispute with the Australian Energy Regulator, adds further time.

The process is governed by Chapter 5 of the National Electricity Rules. Developers and their counsel need to understand the distinction between negotiated and standard connection services, the basis on which augmentation costs are allocated, the circumstances in which a connection offer can be disputed, and the timeframes within which each step must be completed.

Power as a project obligation, not a utility service

The Commonwealth and state governments are now actively shaping the conditions on which data centre investment will proceed. On 23 March 2026 the Australian Government released its Expectations of data centres and AI infrastructure developers (Expectations): five national expectations issued as a commitment under the National AI Plan, applying to new and expanded developments.

The Expectations do not create, vary or extinguish any legal obligation. Their force is procedural: energy-intensive proposals not closely aligned with them will not be prioritised in Commonwealth regulatory assessments. That distinction matters for advisers. Alignment is not a compliance question with a defined standard and a defined penalty. It is a positioning question that affects how quickly and how favourably a proposal moves through approval processes, and it will need to be evidenced rather than asserted.

The second expectation, directed at Australia’s energy transition, reframes power as something a project must bring rather than something the grid supplies. Developers are expected to:

The policy objective underlying each of these is that new data centre load should not raise energy prices for other consumers. Structuring a project so that it can demonstrate this (in contracts, in cost allocation and in the evidence put before regulators) is a legal exercise as much as a commercial one.

Power purchase agreements and latent legal risk

Renewable procurement and emissions commitments are increasingly reflected in hyperscaler tender requirements, service agreements and project financing assumptions. Together with the Commonwealth expectations, this pushes developers toward long-term power purchase agreements with renewable generators. PPA structuring carries significant latent legal risk.

Two exposures recur. The first is basis risk: a mismatch between the PPA’s financial settlement point and the data centre’s physical supply point can leave a developer exposed to electricity cost movements it assumed were hedged. The second is certificate integrity: large-scale Generation Certificates created under the PPA must be generated, surrendered and reported in a manner that satisfies both the regulatory framework and the developer’s customer contractual commitments, which are not always drafted to the same standard.

The drafting of force majeure and curtailment provisions is a further area where template agreements frequently fail to reflect commercial intent. Specialist energy advice is required at the heads of agreement stage, not after term sheets are signed.

Water: from operational input to condition of approval

Water for cooling has moved from an operational detail to a matter of public and political attention, and it is now a live approvals issue.

The third Commonwealth Expectation addresses sustainable and efficient water use. Data centres are expected to minimise water use through innovative and efficient solutions, to engage early with water utilities, communities and First Nations peoples in determining appropriate locations and secure water sources, to adopt efficient cooling technologies, and to use non-potable water and pursue circular water opportunities where possible. Operators are also expected to:

New South Wales has released policy frameworks setting expectations that proponents implement world-class energy, water and environmental standards, including funding additional water supply, ensuring no net cost to consumers and communities, and applying world-class efficiency standards. Other states are moving in similar directions.

The practical effect is twofold. These instruments raise the upfront and ongoing cost base of development. They also supply local communities and public interest groups with an authoritative reference point against which to frame objections. A proposal that cannot demonstrate alignment on water should expect that gap to be identified during consultation, and to be reflected in approval timeframes.

Fibre: connectivity, carrier rights and the infrastructure gap

The carrier framework

Connectivity is a foundational constraint and the one that receives the least legal attention at early project stages. Data centres require diverse, high-capacity fibre connections from multiple carriers, and the framework governing telecommunications infrastructure in Australia is complex.
Under the Telecommunications Act 1997 (Cth), carriers hold statutory powers facilitating the installation of network infrastructure on public and, subject to conditions, private land. Those powers are significant but not unlimited, and carriers who fail to comply with notification and consultation requirements may find their works open to challenge.

For developers, the carrier framework creates a two-sided exposure. As a landowner whose site a carrier seeks to access, the developer has rights to negotiate access and may seek compensation. As a party procuring connectivity, it has rights to impose delivery obligations on carriers and to seek remedies for failure to deliver. Both sides require active legal management.

Dark fibre, IRUs and conduit rights

Most hyperscale and colocation facilities procure dark fibre (unlit capacity over which the operator or its customers run their own equipment) through long-term indefeasible right of use agreements (IRU). IRUs are typically structured as 20 to 30 year licences over specified fibre strands. The legal issues in negotiation include the characterisation of the right as a licence or a profit à prendre, the treatment of the agreement on carrier insolvency, the scope of maintenance obligations and the allocation of risk for fibre cuts or degradation, and the enforceability of diversity and redundancy commitments.

These questions sit at the intersection of property law, telecommunications regulation and infrastructure contracting. Developers who rely on standard-form or lightly negotiated IRUs frequently discover, at the point of an outage or an insolvency, that their agreements do not provide the protections they assumed.

Wayleaves, route diversity and the planning intersection

Genuine route diversity requires fibre to enter a campus by physically separate routes across land in multiple ownerships. Wayleave agreements with councils, state road and rail authorities and private landowners must be negotiated individually, with terms covering maintenance access, upgrade rights and treatment on sale of the land.

The intersection between fibre infrastructure and the development approval process is also underappreciated. Conditions requiring the undergrounding of communications infrastructure, the dedication of conduit routes or the provision of carrier access points within a development are increasingly common in major consents, and can affect site layout, construction cost and program if they are not anticipated at the design stage.

The operating layer: critical infrastructure security

Land, power, water and fibre determine whether a facility can be built. A fifth framework determines how it must be operated, and it bears on design decisions long before completion.

Under the Security of Critical Infrastructure Act 2018 (Cth), the storage and processing of data is a regulated sector, and a facility can be a critical data storage or processing asset, particularly where it holds or processes business critical data for the Commonwealth or for operators of other critical infrastructure. Characterisation is not always obvious, and a colocation operator may hold critical assets on behalf of its customers without having turned its mind to the question.

Where the Act applies, the obligations are positive and ongoing. They include registering the asset, adopting and maintaining a risk management program addressing cyber, personnel, physical and supply chain hazards, reporting annually on that program, and mandatory reporting of cyber security incidents to the Australian Signals Directorate within strict timeframes (12 hours for a significant impact and 72 hours for a relevant impact). Assets declared to be systems of national significance attract enhanced obligations, and government assistance and information-gathering powers sit above the regime.

For developers and tenants, the practical point is that critical infrastructure compliance is a design and governance input, not a post-completion afterthought. Security obligations, customer flow-downs and incident response responsibilities should be allocated in build and colocation contracts from the outset.

The compounding problem

Each constraint is legally complex in isolation. The compounding challenge is that the timelines for resolving them do not align, and progress on one is frequently contingent on progress on another.

A grid connection offer cannot be finalised without a confirmed site. A site cannot be acquired with confidence without visibility on power and connectivity. Carriers will not commit infrastructure without confidence in the tenant’s operational timeline, which depends on planning approval, which in turn depends on demonstrated servicing. Energy and water alignment now bears on how quickly the approval pathway moves at all.

Breaking that interdependency requires an understanding of each constraint and how to sequence the legal and regulatory workstreams accordingly. It also requires developers to accept that expenditure (on connection applications, options over fibre routes, pre-lodgement planning engagement) must be committed before full site certainty is achieved.

What does this mean for developers and investors?

Australia’s AI capacity will be determined less by the pace of the technology than by the legal pathways available to site, power, cool and connect it.

Thomsons advises developers, operators, investors and government on the delivery of data centre and digital infrastructure projects, across site acquisition and planning, grid connection and energy procurement, water and environmental approvals, telecommunications and fibre contracting, and critical infrastructure compliance.