The need to increase data-centre sustainability has become a central topic as demand for digital infrastructure grows across Australia.
The term “data centre” may seem innocuous enough. You might picture a plain, low-profile building, often outside a major city. But there’s much more to it. It takes a tremendous amount of infrastructure to build and operate the data centres that power AI, handle massive datasets and support the digital infrastructure of the future.
AI’s computational demands are reshaping the data-centre landscape. In Australia, data centres are now among the fastest-growing sources of electricity demand in both the National Electricity Market and the Wholesale Electricity Market, according to the Australian Energy Market Operator (AEMO). EY has forecast Australian data-centre energy demand could rise from approximately 4 terawatt-hours (TWh) in 2025 to 21.4TWh by 2035, fuelled by AI and hyperscale cloud growth.
And the amount of power required is growing quickly. For Australia’s energy sector, that creates a new planning challenge: how to connect large, continuous loads while the grid is already managing coal retirements, renewable energy growth, storage investment and transmission constraints.
To keep data centres from overheating, many owners rely on evaporative cooling systems, which use water to remove heat through evaporation. While effective, this method can consume vast amounts of water. In Australia, where water security is a persistent concern, especially during drought and in fast-growing urban corridors, this level of consumption places increasing pressure on local communities.
Want more content like this? Subscribe to Energy and discover all you need to know about the energy transition.
As AI growth marches forward, how can we mitigate this drain on resources?
One opportunity is improving efficiency inside the data centre itself. As AI models grow in complexity and size, running them efficiently requires architectural and software advances. Traditional data centres, built around general-purpose central processing units (CPUs), are no longer sufficient for many modern AI workloads.
Alternatives include general processing units (GPUs), which can execute thousands of operations in parallel, making them ideal for intensive AI workloads and helping improve performance per watt.
There is also the opportunity to make the data-centre building, GPU and racks interdependent for sustainable gain, including through higher-density systems, liquid cooling and designs that reduce energy and water use.
As AI demand accelerates, so does the need for energy, water, building materials and land. Instead of starting from scratch, many forward-looking companies are finding opportunities in what already exists by repurposing or co-locating with infrastructure built for other purposes. In Australia, that could mean siting facilities near renewable energy zones, existing substations, recycled-water infrastructure or industrial precincts.
Community engagement is also critical. Every new data centre must go through a complex approval process involving regulators, elected officials and community members. These stakeholders are often concerned with environmental impact, infrastructure strain and long-term public benefit.
For project teams, digital design and planning tools have an important role to play here. Autodesk works with architecture, engineering, construction and operations teams to help model, coordinate and assess infrastructure projects earlier in the process, supporting more informed decisions around energy use, materials, water, site impact and long-term performance.
Building and operating data centres sustainably is a complex systems problem. While it cannot be solved by one solution, there are many opportunities to work in tandem to minimise the environmental impact of AI.
For Australia, the opportunity is to shape this new infrastructure so it supports digital growth, grid reliability and a more sustainable energy future.
Learn more at autodesk.com.au
This piece was originally published on the Autodesk website.
