Australia’s data centre sector is set to become a significantly larger consumer of electricity over the next decade, with the Australian Energy Market Operator (AEMO) forecasting annual consumption to rise from around 5TWh today to 34TWh by 2035-36.
The forecast, contained in AEMO’s 2026 Electricity Statement of Opportunities (ESOO), would see data centres increase their share of National Electricity Market (NEM) operational consumption from around 3% today to 13% within a decade. AEMO has for the first time included additional data centre analysis alongside the ESOO, reflecting the sector’s increasingly significant role in demand forecasting and power-system planning.
AEMO chief executive Daniel Westerman (above) said the digital economy was becoming a core part of Australia’s future energy requirements. The 2026 ESOO estimates that around 165 data centres currently consume approximately 5TWh of electricity a year. Average monthly data centre demand across the NEM has risen from almost 200MW in 2018 to more than 700MW in June 2026.
New South Wales currently accounts for 66% of data centre consumption, followed by Victoria at 30%, with South Australia at 2%, Queensland at 1% and Tasmania below 1%. Western Australia sits outside the NEM.
67GW pipeline, but significant attrition
The scale of the development pipeline is considerably larger than the forecast consumption growth. Network service providers reported 225 data centre projects at various stages of the connection process for the 2026 ESOO, with combined proposed connection capacity of 67GW, up from 38GW in last year’s assessment. However, AEMO cautions against treating that figure as a forecast of future demand.
Around 36% of the projects included in the 2025 ESOO project list have subsequently been cancelled, including more than 30% of projects previously classified as committed. AEMO says the most advanced developments are concentrated in NSW and Victoria, while projects in South Australia and Tasmania are generally less mature.
AEMO has incorporated that uncertainty into its forecasting methodology, applying probability weightings to individual projects based on factors including project and connection status and developer size. It then combines the resulting project-level forecast with an economic forecast that considers broader digitalisation, AI adoption and Australia’s potential role as an Asia-Pacific data centre hub.
The forecasting methodology also assumes data centres take time to reach full utilisation. AEMO says operators commonly indicate a five-to-10-year period to reach capacity. Current operating data centres remain well below that level, averaging around 27% utilisation in FY26, but AEMO projects that a mature fleet, allowing for planned capacity redundancy and diversified compute profiles, will average 45-51% utilisation depending on data centre type.
34TWh central forecast, 52TWh high-growth case
Under AEMO’s Step Change scenario, data centre consumption is forecast to triple to around 15TWh by 2029-30, representing 7.8% of NEM operational consumption, before reaching around 34TWh, or 13%, by 2035-36. But AEMO has also tested a higher-growth scenario to examine what would happen if data centre demand develops more rapidly than assumed in its central forecast.
Under the High Data Centre Growth sensitivity, annual consumption reaches around 52TWh by 2035-36, which is 18TWh, or 56%, above the Step Change forecast. AEMO says the additional growth is concentrated particularly from the early 2030s.
The sensitivity is significant because AEMO has kept the supply-side assumptions unchanged. In other words, it is testing what happens to reliability if data centre demand grows faster without automatically assuming that additional generation, storage and transmission will arrive alongside it.
The result is a substantially tighter reliability outlook. In the high-growth case, expected unserved energy in 2035-36 rises to 5.58% in NSW, 3.95% in Victoria, 2.10% in South Australia and 1.71% in Tasmania, compared with 2.24%, 1.82%, 0.81% and 0.31% respectively under the Government Schemes and Actionable Developments assessment. Queensland is largely unaffected in the sensitivity because data centres currently account for relatively little of its forecast demand growth.
AEMO says that if the higher level of data centre growth assumed for 2035-36 were instead reached four years earlier, in 2031-32, additional investment in generation, storage and transmission would be required from 2032-33 in NSW, Victoria and South Australia to maintain comparable reliability outcomes.
Supply will need to keep pace
The data centre analysis comes against a broader warning from AEMO that the reliability outlook depends heavily on new generation and storage projects actually being delivered. Under the ESOO’s Government Schemes and Actionable Developments assessment, reliability is maintained within the relevant standard until 2033-34 in NSW, Victoria and South Australia and 2034-35 in Tasmania.
However, when AEMO restricts the assessment to projects that meet its more stringent committed and anticipated development criteria, reliability gaps emerge earlier: from 2030-31 in NSW and Victoria, 2031-32 in South Australia, 2032-33 in Queensland and 2033-34 in Tasmania. The distinction is important for data centre developers because the sector is competing for the same generation, storage and network investment required to replace retiring coal generation and accommodate broader electrification.
AEMO says data centre growth is now a key driver of electricity consumption particularly in NSW, Victoria and South Australia. Its high-growth analysis concludes that timely investment in generation, storage and supporting transmission will be required if demand develops faster than the central forecast.
Data centres’ role in managing demand
AEMO also identifies a potential opportunity for data centres to provide more flexibility than before from the sector. Greater demand flexibility during periods of supply scarcity could materially reduce the investment required in the electricity system, AEMO says, adding that data centres “may provide greater load flexibility than has been historically observed across existing facilities.”
That could become increasingly relevant as policymakers consider how large new electricity users should integrate with the NEM.
AEMO notes the federal government’s proposed requirements for future large data centres to underwrite their own power supply, as well as the Australian Energy Market Commission’s (AEMC) July 2026 advice to the Energy and Climate Change Ministerial Council, which recommended driving new renewables through the Renewable Electricity Guarantee of Origin scheme, introducing a data centre contracting obligation, introducing market registration requirements, and supporting demand flexibility and co-location with generation through connection agreements.
AEMO has not explicitly modelled the effect of those proposed requirements because it cannot identify which specific generation or storage projects data centre developers might ultimately contract with. It says, however, that if the proposed framework is implemented, some of the additional supply investment needed to support higher data centre growth could be delivered directly by the developers, reducing supply scarcity risks for other consumers.
As a result, the 2026 ESOO therefore presents data centres as both a major new source of electricity demand and a factor that needs to be incorporated more explicitly into system planning. While the central forecast points to a substantial increase in consumption, AEMO’s treatment of the 67GW project pipeline also highlights the gap between proposed capacity and demand that is ultimately likely to materialise.