A new line item on the P&L: why Canberra wants datacentres to pay for their power, and why two jurisdictions pushed back
For CFOs and real‑estate chiefs, location decisions for AI datacentres are becoming a regulatory and infrastructure risk. The federal government is pushing nationally consistent, binding rules that would require new large datacentres to underwrite new power, pay for grid connections, curb water use and improve energy efficiency, and to “put at least as much energy into our grid as they take out of it.” That push provoked public objections from Queensland and, more recently, the Northern Territory.
The scale that got everyone’s attention
The Clean Energy Finance Corporation and consultants Baringa modelled the issue in December 2025. Their report projects aggregate data‑centre capacity in Australia rising from a 2025 baseline of about 1.35 GW to between 4.7 GW and 7.4 GW by 2035 under different growth scenarios (CEFC/Baringa, Dec 2025). In the study’s high‑growth case, datacentres could represent as much as roughly 11% of national electricity consumption by 2035.
The CEFC/Baringa modelling is explicit about assumptions. The GW figures represent aggregate nameplate capacity for IT load as modelled across existing and proposed facilities, and the recommended mitigation, roughly 3.2 GW of additional renewable generation plus about 1.9 GW of battery storage by 2035, refers to installed capacity required under the report’s scenario assumptions to limit wholesale‑price and emissions impacts. Those findings are model outputs that depend on location, timing, commissioning sequences and assumed power‑use effectiveness (PUE) and workload growth rates.
What the federal plan would do (and what that language actually means)
Prime Minister Anthony Albanese has framed new rules so that large datacentre projects carry a “legal obligation” to underwrite additional power supply, meet their connection costs and offset consumption with verified clean generation (statement, July 2026). Put plainly, the government is aiming for annual MWh matching and demonstrable additionality, not literal second‑by‑second export parity at the local substation.
“Underwrite new power supply” can take several practical forms. Examples include long‑term contracted new‑build power purchase agreements (PPAs) that finance additional wind or solar, developer financing of new generation assets, or commitments to procure firming capacity, such as batteries, pumped hydro or other dispatchable options tied to commissioning milestones. Policy design will need to be precise about which approaches count as acceptable additionality and how they are audited.
Why Queensland and the Northern Territory said no
At a late‑July 2026 meeting of energy ministers, Queensland’s treasurer and energy minister David Janetzki rejected parts of the federal approach, calling the proposals “underdeveloped” and warning they would cede local control to Canberra. He said: “Queensland will always support proposals that deliver affordable, reliable and sustainable power, however, we will not support underdeveloped ideas that hand increased power to Canberra at the expense of Queenslanders.” The Northern Territory followed suit, emphasising the Territory’s unique grid and cost profile.
The political dispute is not just rhetorical. States worry that heavy federal prescription, especially if tied to connection rules or national market powers, could influence where investment flows, who pays for network upgrades, and what projects get approved first. Canberra argues a national rule set is needed to protect consumers from price shocks and to ensure emissions‑aligned growth of digital infrastructure.
The timing mismatch that worries system planners
A recurring technical problem is tempo. Datacentre projects can be developed and brought online in roughly 18-24 months, while transmission upgrades and new utility‑scale renewable projects commonly take three to five years or longer to plan, approve and commission. That timing gap can leave the grid short of firm capacity in the short term, potentially pushing up wholesale prices and stressing local networks unless developers commit to genuinely additional firming or approvals are staged to match supply build‑out.
Policymakers therefore face three hard design questions: what size or threshold of datacentre triggers the rules, what counts as verifiable additional generation and firming, and how to speed or prioritise network investment for likely cluster locations.
Industry and community reactions
Industry bodies point out that many operators already secure renewable supply. Data Centres Australia’s CEO Belinda Dennett told ABC News in July 2026 that operating facilities in the market were already offsetting a large share of their energy use through contracts and projects. Still, industry offsetting does not automatically resolve system‑level timing, locational or network issues, a PPA that finances generation hundreds of kilometres away may not ease congestion or local price impacts in the short term.
Community concern is palpable. Residents and local planners frequently raise questions about local water use for cooling, noise, and diesel generator emissions used for backup. There are practical engineering responses, such as air‑cooled designs, recycled water, closed‑loop cooling and onsite storage, but they change cost and efficiency trade‑offs and must be considered at the approvals stage.
How enforcement, definitions and legal levers might work
There are several possible enforcement and legal routes. The Australian Energy Market Operator (AEMO) sets market and connection rules. AEMO operates the wholesale market, manages system planning and advises on technical connection requirements. The federal government could seek AEMC rule changes, use AEMO’s planning instruments, or legislate national standards that set thresholds and compliance paths for large loads. The exact legal mechanism has not been publicly specified.
Verification options include independent auditors and registries, commissioning‑tied certificates of additionality, and linking connection approvals to proof that new generation or firming is under construction or contracted. A weakly defined “match the grid” requirement would invite both legal challenges and creative accounting, so precise, auditable metrics are essential.
Practical choices for executives and investors
- Energy and infrastructure directors: produce a site exposure map that lists peak MW, expected in‑service date and connection constraints for each location. Run a sensitivity analysis on wholesale prices under scenarios where data‑centre demand arrives ahead of planned renewables and transmission (for example: +1 GW, +3 GW, +5 GW).
- Real‑estate and transactions teams: require auditable additionality commitments from developers, PPAs that finance new build, developer equity into generation, or firming contracts tied to milestones, before greenlighting sites.
- Investors: ask for water‑use plans and contingency cooling strategies in due diligence. Where applicable, require measurable targets (litres per MWh or similar) and independent verification.
Vignette: a hyperscaler choosing between a Queensland site with cheap land but no committed local renewables pipeline, and a state that has banked transmission upgrades and firmed renewable projects, will need to price in the cost and timing of delivering additional firming or face slower approvals and higher connection tariffs.
Trade‑offs and policy pathways
The choices reduce to three broad options:
- Strict national mandates that bind developers to additionality and connection costs. Clear on paper, but politically contested and potentially deterrent if thresholds and compliance paths are clumsy.
- Market‑led approaches relying on corporate PPAs and voluntary standards. Faster and less prescriptive, but they risk uneven outcomes and localized grid stress.
- Coordinated, conditional approvals that tie datacentre commissioning to funded transmission projects or to verified build‑out of renewables and storage. Administratively demanding, but likely the best way to manage sequencing risk.
The CEFC/Baringa modelling gives one concrete policy lever: under its scenario assumptions, adding about 3.2 GW of renewables and around 1.9 GW of battery storage by 2035 would materially reduce the risk of wholesale price rises and neutralise additional emissions from datacentre growth. That is not a guaranteed outcome, it depends on where and when capacity comes online, but it provides a pragmatic target for coordinated investment and planning.
Key questions, and short answers
- Who is opposing the federal renewables mandate for datacentres?
Queensland and the Northern Territory have publicly rejected parts of the federal plan, calling sections underdeveloped and warning against ceding local decision‑making to Canberra (statements at late‑July 2026 energy minister meetings). - How large could datacentre demand become?
CEFC/Baringa (Dec 2025) modelled aggregate IT‑load capacity rising from about 1.35 GW in 2025 to between 4.7 GW and 7.4 GW by 2035; in its high‑growth scenario datacentres could represent roughly 11% of national electricity consumption by 2035. These are model outputs that depend on location, PUE and commissioning timing. - What does “underwrite new power supply” mean in practice?
It typically means developers must guarantee additional generation or firming capacity via new‑build PPAs, direct investment in generation assets, or contracted firming (batteries, pumped hydro, etc.) that are verifiably additional to the existing pipeline. - Why do planners worry about timing?
Datacentres can be built in about 18-24 months; major transmission and utility‑scale renewables often require three to five years. If many large loads arrive before firm capacity and network upgrades are ready, local wholesale prices and reliability can be affected. - What should my organisation do this quarter?
Map exposure: quantify peak MW per site, stress‑test wholesale price exposure under plausible demand arrivals, and require auditable additionality from developers. Those three steps turn a policy risk into a manageable commercial metric.
Where this goes next, and why it matters
The debate is about sequencing more than ideology. Australia’s ambitions to host AI and cloud capacity are a genuine economic opportunity. Getting the timing and rules right will determine whether that expansion happens with constrained grid costs and community support, or amid higher consumer bills, local backlash and legal disputes.
Policymakers still need to define thresholds, measurement standards for additionality, and enforcement mechanisms, and states want assurances on local cost and control. For businesses, the immediate priority is operational: treat location as a regulatory risk, demand verifiable commitments on additionality and water use, and bake connection and timing risk into project economics.
CEFC/Baringa (Dec 2025): the modelling provides a clear, conditional target, roughly 3.2 GW of new renewables and 1.9 GW of battery storage by 2035 under the scenarios modelled, to help avoid material price and emissions impacts from datacentre growth.