AI is changing the world order. Now is our time to shape it.

Register for the AI Leadership Summit in Sydney | 27 October

Opinion article

Nine governments, one rulebook, few details: why Australia's data centre framework has some way to go

National standards for data centres are a welcome step, but without clear rules on energy, water, land use and state approvals, Australia risks replacing regulatory fragmentation with a new layer of uncertainty.

Australia is racing to build the regulatory and policy framework for data centres. However, a fragmented approval system and a lack of regulatory clarity risk impeding capital deployment in the sector, unless regulators get the rulebook right.

On 26 August 2026, every Australian Government agreed to develop consistent mandatory national standards for large data centres covering energy, water and land use – nine governments, one rulebook. 

Getting every state and territory to align on a single framework for one asset class is a significant outcome for Australian federalism. It reflects welcome recognition that data centres are now critical economic infrastructure rather than just commercial real estate. 

However, the announcement is short on detail. Two key challenges will decide whether this framework works in practice or simply adds friction to a sector that requires speed and certainty.

One rulebook requires a streamlined approval system

Commonwealth standards on energy, water and land use are designed to sit alongside state planning and environmental approvals, not replace them. In practice, every large data centre proponent must navigate multiple regulatory schemes – both Commonwealth and state – applying different thresholds, timeframes and degrees of discretion, with no clear mechanism yet for determining how conflicts between schemes would be resolved. 

The Commonwealth standards are intended to set nationally consistent minimum requirements for data centres and facilitate assessments and approvals across various regulatory schemes – providing faster regulatory processes and simplified dealings with regulators. 

This has the potential to streamline delivery while strengthening the sector’s “social licence to operate”. But its success will depend heavily on the timely delivery of national standards. 

NSW isn't waiting for the Commonwealth to sort this out. It is already planning to legislate compliance as a condition of grid connection, while its transmission operator has implemented a policy to allocate transmission capacity ahead of the national framework. If other states follow with their own versions before the Commonwealth standards are finalised, Australia will fail to achieve a unified rulebook in practice.

Victoria has also announced its own requirements for new data centres, including renewable energy and storage obligations, residential buffer zones and local investment contributions. Regulatory duplication can extend approval timelines, reduces outcome certainty and makes Australia less competitive just as global AI infrastructure investment is being deployed across Asia, the Gulf and North America. For National Cabinet to deliver on its promise, Commonwealth standards must resolve key issues upfront rather than leaving them to be settled project-by-project.

Vague concepts, unstable standards

The second challenge is that the Commonwealth has invoked “national interest” and the need for data centres to earn a “social licence to operate” to justify intervention, but has yet to define what the proposed standards mean in practice. Deals stall not only under strict criteria, but under ambiguous ones; owners can’t commit capital against a moving target. Indistinct standards shift uncertainty onto developers. 

Energy and water illustrate how this plays out. Data centres used for real-time AI inference need continuous power with little tolerance for interruption. Yet, government also expects them to act as “net generators”, adding power to the grid, paying connection costs, and shifting load to help stabilise the system under strain. Operating as a guaranteed baseload customer while functioning as a flexible grid asset are structurally different roles, and government hasn't specified how the economics of combining those roles will work. 

Water follows a similar pattern – operators will need to minimise usage, fund additional infrastructure, and meet efficiency thresholds that remain undefined.

Getting either setting wrong in one direction sees hyperscale demand bolted onto constrained infrastructure without matching investment, leaving households to absorb the cost. Getting it wrong in the other direction imposes substantial underwriting obligations, including long-term offtake agreements, transmission contributions, and water levies, that are too onerous to finance at the pace needed to compete globally.

What is needed is a single, stable methodology: defined formulas for generation to be added; explicit compliance mechanisms such as offtake agreements; and equivalent clarity on water thresholds and cost-sharing. Developers can price almost any fixed obligation – they can't effectively finance one that is subject to change and lacks clarity.

 Land use faces a similar problem from the opposite direction; there is currently no standard at all. Data centres are competing directly with housing and community infrastructure for suitable sites near power and connectivity in corridors where states are accelerating housing supply. Government wants facilities located “appropriately” with “genuine community engagement”, but has defined neither, nor specified who arbitrates when a council and State planning authority disagree.

Clear siting rules are particularly important where data centres compete with housing and community infrastructure for well-connected land. Poorly located projects could undermine the community support which governments are seeking to build, while requirements introduced only after capital has been committed risk disrupting projects which have been designed, financed or developed under earlier settings. 

Contested projects risk forcing these definitions through the courts, eroding public trust and delaying critical projects.  

Getting this right

Done properly, a genuinely harmonised national standard would be one of the most valuable pieces of policy infrastructure Australia builds this decade, letting capital move with confidence instead of navigating nine separate approval systems. That value depends on predictability, clarity and precision, delivered promptly.

Before legislating in 2027, government needs to resolve how Commonwealth standards interact with state approvals when they conflict and attach stable, explicit criteria to “national interest” and “social licence”, rather than leaving them open to renegotiation. Industry shouldn't wait for those answers; energy, site selection and financing models should already be stress-tested against potential outcomes and FIRB implications managed early. 

The word “one” in “one rulebook” is carrying a lot of weight. Australia has months, not years, to make sure it holds up.

CEDA Members contribute to our collective impact by engaging in conversations that are crucial to achieving long-term prosperity for all Australians. Find out more about becoming a member, our ESG and AI Communities of Best Practice or getting involved in our research today.
About the authors
SM

Samy Mansour

See all articles
Samy Mansour specialises in investments, joint ventures and major transactions across the energy, resources, logistics, infrastructure and digital sectors. 
NT

Nick Thomas

See all articles
Nick Thomas specialises in environment, planning, sustainability and major project delivery, with a focus on the energy, resources, infrastructure and property development sectors. 
AD

Alex Danne

See all articles
Alex Danne is Head of Energy at Clayton Utz and a leading energy and infrastructure lawyer.