Data Centres Energy Demand Outlook in Australia

This report presents Oxford Economics Australia’s and the Australian Energy Market Operator’s joint forecasts of electricity consumption by data centres in Australia. It provides a forward-looking assessment of how data centre electricity consumption is likely to evolve.

Headline research findings

• Under the central scenario, Step Change, data centre electricity consumption across the NEM is forecast to increase from 5.1 TWh in FY26 (5.2 TWh nationally) to 15.5 TWh (15.8 TWh nationally) by FY30 and 33.7 TWh (34.3 TWh nationally) by FY36.

• The near-term outlook is primarily driven by the ramp-up of existing data centres and likely expansions of existing sites. These facilities form a large base of connected and committed capacity and are still moving toward mature operating levels.

• After FY30, new data centres become the main driver of growth in energy demand as currently prospective projects move through the project pipeline, connect to the grid and begin to ramp up their demand. Due to the inherent uncertainty of how many of these projects progress through planning, commercial and delivery hurdles, uncertainty in the forecasts increases materially from FY30.

• Demand remains concentrated in New South Wales and Victoria, with the two regions accounting for over 85% of NEM data centre electricity consumption to FY36. The concentration of data centres in New South Wales and Victoria has implications for the grid. Data centres are forecast to consume around 12% of total state grid-supplied electricity in NSW and VIC by FY30, rising to 19% by FY36.

• The 2026 forecasts are a material upgrade from 2025. Under Step Change, energy demand is 31% higher by FY30 and 47% by FY36, before moderating to 27% in the long run by FY50. The upgrade reflects the larger development pipeline, with connection requests significantly higher than in 2025 in MW terms alongside updated core assumptions informed by recent data releases.

• The inherent uncertainty over the pace and scale of this emerging industry is reflected in AEMO’s Accelerated Transition and Slower Growth scenarios.

• Accelerated Transition reflects a stronger development cycle, with faster cloud adoption, stronger AI demand, greater servicing of international workloads and a larger share of prospective developments progressingthrough the development pipeline.Thislifts NEM demand to 20.1 TWh (20.4 TWh nationally) by FY30 and 52.5 TWh (53.5 TWh nationally) by FY36, with Victorian and transmission-connected developments playing a larger role.

• Data centre demand continues to grow even under Slower Growth, reaching 10.5 TWh (10.8 TWh nationally) by FY30 and 15.1 TWh (15.7 TWh nationally) by FY36, as existing facilities, likely expansions and well progressed prospective projects continue to ramp with a reduced need for large-scale new development.

Model refinements

Three key model refinements have been made in the 2026 forecast update to incorporate additional data and evidence as the Australian data centre market develops.

1. Probabilistic weighting was introduced in the Project Approach to provide a more nuanced estimate of the likelihood of prospective projects materialising into future connected capacity. Some projects may be withdrawn, duplicated across multiple developers responding to the same underlying demand, or lodged before commercial, planning and connection hurdles have been resolved. Rather than treating each prospective project as likely or unlikely, the updated approach assigns projects a probability of progression. These probabilities are statistically determined by observed changes in the connection request database between 2025 and 2026, including characteristics like project status, connection status, localised competition and whether the proposed project timeline is realistic. This provides a more structured way to account for potential phantom demand while integrating information from current market activity to inform the outlook.

2. A logistic curve has been adopted to forecast the intensity of electricity consumption in the Economic Approach to better reflect the relationship between the consumption of digital services and electricity consumption. The previous approach used a linear trend to estimate electricity consumption per dollar spent on web and data services. Recent data points suggest this relationship is better represented by a logistic curve reflecting faster short-term growth driven by compositional shifts from traditional web and data services toward more electricity intensive cloud, AI and data processing activity which will settle as the market matures.

3. An international demand component is introduced to the Economic Approach to better capture Australia’s potential as a data centre hub in the APAC region servicing international workloads. Public data on international workloads is extremely limited. To estimate the potential scale of this effect, the model compares Australia’s installed data centre capacity relative to the size of its economy with the equivalent ratio for Asia Pacific excluding China and Australia. These estimates suggest that around 5% to 10% of current capacity could be servicing international demand, equivalent to around 1% to 1.5% of APAC consumption (excluding China and Australia).

We expect to continue refining the model methodology over time as the data landscape improves. The probabilistic project weighting will become more robust as additional years of data are collected, providing a larger sample size for analysis. Similarly, electricity intensity and international servicing demand remain areas where public data is limited, but market conditions are changing quickly. As more data becomes available, we will continue to refine the forecasts and reduce the uncertainty in the forecasts.