Australian government sets national expectations for data centres

March 23, 2026 at 7:37 AM GMT+8

The Australian government has introduced a new set of national expectations for data centres and AI infrastructure developers, signalling a more interventionist approach to managing the rapid expansion of energy-intensive digital infrastructure while seeking to align investment with national priorities.

Announced as part of the government’s National AI Plan, the framework outlines five core expectations covering national interest, energy, water, workforce, and innovation. While not formal regulation, the guidance will be used to prioritise projects through Commonwealth processes, effectively shaping which developments progress fastest and creating a de facto filter for new proposals entering the approval pipeline.

Industry minister Tim Ayres (above) said the move reflects a broader push to ensure AI infrastructure delivers domestic benefits. “Australia is open for business – but the kind of business that puts Australia’s national interest first,” he said.

At the centre of the framework is a stronger emphasis on “social licence”, with developers expected to demonstrate how projects contribute to economic growth, data sovereignty and community outcomes. This includes requirements to protect sensitive data, engage with local communities, and minimise negative impacts.

Renewable generation

Energy has emerged as a defining theme. The government expects new data centres to underwrite additional renewable generation, cover their full share of transmission and distribution costs so these are not passed on to households or businesses, and participate in demand flexibility to support grid stability. The framework also makes clear that speculative or underdeveloped proposals that congest approval pathways will not be prioritised, signalling a tighter approach to how projects are assessed.

Energy minister Chris Bowen said the aim is to ensure infrastructure growth does not drive up costs for consumers. “Data centres have great potential to support our grid and expand new renewable investment, but it’s important we work together… to keep our system secure and energy prices low,” he said.

Water use is also addressed explicitly, with operators expected to adopt efficient cooling technologies, prioritise non-potable water sources where possible, and provide transparent reporting. The framework also calls for investment in local skills and supply chains, alongside expectations that hyperscalers and emerging “AI factory” operators make compute capacity available to Australian startups and researchers on favourable terms.

Targeting larger facilities

The government said proposals most closely aligned with the expectations would be prioritised, while those that do not meet the criteria may face slower regulatory progress. The framework applies to large-scale facilities including hyperscale, colocation, and AI-focused deployments, but excludes smaller edge and on-premises systems.

Data Centres Australia, the industry’s peak body, broadly welcomed the direction of travel but highlighted areas of concern around scope, implementation, and regulatory clarity. Chief executive Belinda Dennett said many of the expectations are already being met by operators. “Many of the expectations outlined by the government are already being met by Data Centres Australia members – voluntarily and without regulatory obligation,” she argued, adding that the industry supports measures to raise standards across the market.

The association pointed to existing investment and operational practices as evidence of this, noting that data centre operators have invested AUD 3.1 billion in energy infrastructure since 2020, with a further AUD 7.2 billion committed by 2030. It also claimed around 70% of sector energy use is already offset through renewable procurement mechanisms, contributing an estimated 1.5TWh of additional renewable generation.

On water, the group said data centres account for approximately 0.04% of national usage – albeit a current figure rather than a forecast – while highlighting ongoing efforts to adopt low-water and non-potable cooling solutions. It also pointed to existing regulatory obligations under Australia’s Security of Critical Infrastructure (SOCI) Act as evidence that security expectations are already well defined.

However, Data Centres Australia argued that excluding on-premises enterprise compute from the framework creates an uneven policy outcome. While not central to its response, the group noted that on-premises infrastructure still accounts for a significant share of compute capacity and is typically less energy efficient than purpose-built facilities.

Assessment frameworks

Dennett also raised concerns about how the expectations will be applied in practice. “The expectations do not specify how alignment will be assessed or against what criteria proposals will be measured,” she said, calling for clearer and more consistent assessment frameworks to provide certainty for investors.

Energy flexibility requirements were another area flagged for further work. While supporting the intent, the association said current environmental regulations limit the ability of operators to use backup generation or battery systems for grid support, potentially constraining participation in demand response schemes.

The expectation that hyperscalers provide compute access on favourable terms also prompted questions around the role of government in commercial arrangements, with the industry calling for further consultation to avoid unintended impacts on existing programmes.

Despite these concerns, both government and industry appear aligned on the broader direction: that data centre growth will continue, but under closer scrutiny as infrastructure becomes increasingly central to energy systems, economic policy and national capability.

The proposed framework reflects a shift seen in other markets, where governments are moving to define the conditions where hyperscale and AI infrastructure can expand. In Australia’s case, the government’s emphasis is on ensuring that growth in AI and cloud capacity translates into domestic economic, environmental, and technological outcomes, rather than simply accommodating global demand.