Executive summary

Australia’s electricity system is undergoing a profound transition as it moves from a centralised, dispatchable generation model to one incorporating record volumes of renewables. This shift is accompanied by increased demand from data centres, electrification, and consumer technologies. Artificial intelligence (AI) is identified as both a driver of this demand and a potential solution. While AI increases electricity consumption via data centres, it is also a tool capable of unlocking material efficiencies and capacity from existing infrastructure without the delays associated with building new transmission and generation. Adoption in Australia currently remains incremental, and the report identifies three primary barriers: risk aversion and weak regulatory signals, a revenue determination process with a strong capital expenditure (CAPEX) bias, and highly siloed data assets.

AI will be a vital unlock as Australia shifts towards the electricity system of the future

The National Electricity Market (NEM) is experiencing a significant rise in consumption, particularly from large industrial loads, electric vehicles, and data centres. To meet this demand, Australia requires 6,000 kilometres of new transmission infrastructure. The cost of this build-out has risen three times relative to the Australian Energy Market Operator’s (AEMO) 2024 Integrated System Plan (ISP), potentially costing taxpayers an additional $600 per year with lead times spanning 8 to 12 years. Globally, data centre electricity consumption is forecast to grow 4.4 times between 2020 and 2035, reaching 1,193 TWh. AI-enabled tools can mitigate these pressures by unlocking 175 GW of transmission capacity and providing US$110 billion in annual savings through reduced outages. Widespread adoption in advanced economies could also lead to a 5-10% reduction in overall electricity demand.

The transition requires a shift from a human-driven, static grid to an AI-driven, adaptive system. While current AI use cases are often incremental—such as predictive maintenance for generators or automated customer service for retailers—transformative applications include self-healing autonomous grids and “energy-as-a-service” ecosystems. These advanced applications can coordinate distributed energy resources like rooftop solar and community batteries to defer the need for large-scale network augmentation.

A fundamental shift in the market is required to overcome the cultural, institutional and regulatory barriers to AI

The report highlights that underinvestment in transformative AI is driven by a risk-based lens, where energy is viewed primarily as vulnerable critical infrastructure. Industry leaders often lack a clear blueprint for deployment, hampered by legacy technology and limited skills. Regulatory incentives also disadvantage AI, which typically falls under operational expenditure (OPEX). In Australia, CAPEX accounts for approximately 70% of allowed revenue for network service providers (NSPs), whereas OPEX and software make up only 30%. This creates a bias toward physical assets over more efficient digital solutions. Furthermore, the Australian market is highly concentrated, with only 20 NSPs serving 11 million customers, compared to approximately 3,000 providers in the United States.

Data accessibility remains a critical hurdle. While wholesale market data is shared in 30-minute increments by AEMO, other essential data types—such as low-voltage network data and granular metering data—are held by individual participants with no strong incentives to share. Specific regulations, including Chapter 7 of the National Electricity Rules and the Privacy Act, limit the ability to on-share data, constraining the development of grid-optimising AI solutions.

Government and industry must work together to unlock the virtuous cycle of AI in energy

To scale AI, joint action is required across three sectors. Government bodies, including the Department of Climate Change, Energy, the Environment and Water (DCCEEW) and the Australian Energy Regulator (AER), should conduct a market study on AI’s role and establish a clear national policy position. The report recommends creating a dedicated “AI in Energy Fund” to finance innovation, citing similar mechanisms in California and the United Kingdom. The tech industry should focus on upskilling utility leaders and providing evidence of return on investment (ROI). Meanwhile, utilities are encouraged to participate in pilots and advocate for regulatory reforms, specifically the adoption of a total expenditure (TOTEX) model to remove the existing CAPEX bias.

Appendix

Australia operates two distinct markets: the NEM, covering the eastern and southern states, and the Wholesale Electricity Market (WEM) in Western Australia. The NEM serves roughly 9 million households with a market value of $25 billion, while the WEM serves 1.2 million households with a value of $1.6 billion. Governance is shared across several bodies: the AEMC acts as the rule-maker, the AER and ERA WA serve as regulators, and AEMO operates both markets. Legislative frameworks include the National Electricity Law and the Security of Critical Infrastructure (SOCI) Act, which imposes strict disclosure and security rules on energy assets.