Australia’s data-centre boom is becoming an infrastructure problem

AEMO describes data centres as among the fastest-growing sources of electricity demand, and notes that their scale, concentration and continuous operation require careful management of system strength, voltage stability and transmission investment. That combination — large, clustered, and running flat — is what separates this class of load from the demand growth the grid is used to planning for.

The National Electricity Market gives the clearest read on the pressure. AEMO counts 162 operational data centres nationally, already accounting for around 2% of purchased electricity. Its 2025 outlook projects data-centre consumption growing roughly 25% a year, reaching about 12 TWh, or 6% of grid-supplied electricity, by 2030, and approximately 34 TWh, or 12%, by 2050. By the end of March 2026, 11 projects above 5 MW, representing 5.4 GW of maximum demand, were progressing through transmission connections — roughly 60% of that capacity in New South Wales and 40% in Victoria.

Not every announced project will be built, and few reach their maximum load quickly. A 2024 market snapshot recorded about 1.3 GW of connected data centres drawing only 340 MW, with individual sites typically ramping 50–100 MW per year rather than arriving at full demand. The planning question is therefore not how to accommodate every proposed megawatt. It is how fast that demand becomes utilised, and where it finally concentrates.

Electricity and transmission come first, and the difficulty is delivery rather than generation. Operators cluster around Sydney and Melbourne for connectivity and latency, while generation capacity is usually easier to develop somewhere else. New South Wales expects data-centre electricity use to rise from 5% of grid-supplied energy in 2026 to 11% by 2030, which puts pressure on generation, storage and transmission at once.

Water is the less visible constraint. Cooling loads can weigh heavily on existing water infrastructure, particularly in areas already managing prolonged supply challenges, and limited transparency around future consumption makes that infrastructure difficult to plan against.

Delivery capacity is constrained on its own terms. Transmission equipment, skilled labour, contractors and engineering capacity are already stretched by the broader energy transition, so data-centre development competes for the same scarce inputs rather than drawing on a separate pool.

Beyond infrastructure, the question becomes one of planning and social licence. Noise, emissions from backup generation, water consumption and proximity to residential areas are increasingly live considerations for regulators and developers alike.

None of this makes data-centre growth a net drain. Large, creditworthy users can help underwrite new generation, storage and water infrastructure that would otherwise struggle to reach a final investment decision, and the capacity they fund does not serve them alone. They also sit underneath the broader digital economy the country is relying on for productivity growth.

So the argument is not really about whether Australia should accommodate further growth. It is about how the supporting infrastructure is funded, where it is located, and who carries the cost of delivering it.

The strength of the opportunity depends on whether supporting infrastructure can scale alongside demand without transferring disproportionate cost to households and other users. That is a distributional question as much as an engineering one, and it is decided early, at the point of siting and connection, rather than later.

Better site selection, additional generation and storage, recycled water and clearer allocation of infrastructure cost would move this from a constraint to an investment opportunity. The instruments already exist. What is missing is the discipline of testing a site against them before the commitment is made.

Markham assesses infrastructure readiness before major development and investment decisions are taken. The assessment examines whether a proposed site carries the power, water, connectivity and supporting infrastructure to serve both current and future demand — and says plainly where it does not.

Australian Energy Market Operator, Digital demand surge: preparing Australia’s power systems for the rise of data centres, 2026.

Infrastructure NSW, NSW Data Centre Consultation Paper, 2026.

RenewEconomy, Huge data centres queue to join Australia’s grid, but not where wind and solar industry wants them, 2025.

Ragusa and Crampton, Clicking and Swiping Away: Hidden Implications of Australian Data Center Water Security and Management, Water, 2026.

Australian Financial Review, Runaway energy build-out costs threaten data centre opportunity, 2025.