Let’s brainstorm a few different titles – for example, Australia’s next industrial boom (data centres) has a materials problem; We can power data centres with renewables — but what about the concrete?
Spinifex is an opinion column. If you would like to contribute, contact us to ask for a detailed brief.
Australia has quietly become one of the most attractive places on Earth to build a data centre. Stable politics, access to renewable energy, fast-tracked approvals in states like NSW and Victoria, and proximity to undersea cables have positioned the country as a serious contender for the AI and digital infrastructure boom, especially in the Asia-Pacific.
There are a lot of conversations happening about the environmental costs of this boom for the nation, primarily centred on energy efficiency and water consumption. But the Materials and Embodied Carbon Leaders’ Alliance (MECLA), in its latest report, argue that this construction boom is racing ahead of an often less central critical sustainability question: what’s locked into the concrete, steel and silicon before a single server even boots up?
That’s the main takeaway from The Next Frontier: Reducing Embodied Carbon in Data Centre Development. Global spending on data centres is projected to approach $3 trillion by 2029, with 60 per cent of the growth in power demand between now and 2028 driven by AI workloads alone. Global cloud computing emissions have already overtaken those of commercial aviation. For an industry under that kind of pressure to build fast, embodied carbon – the emissions associated with the manufacture, transport, installation and replacement of materials in the asset – is the often less attended to consideration next to operational metrics like power usage effectiveness (PUE) and water usage effectiveness (WUE).
Historically, that makes sense. The embodied carbon in data centre development doesn’t compare, from a whole of life perspective, to the emissions associated with other aspects of the asset class such as energy. But as grids decarbonise, the share of carbon associated with the materials used in construction will grow.
Data centres are unusual buildings. Compared with an office tower of similar size, they carry a disproportionate embodied carbon load because of dense mechanical, electrical and plumbing (MEP) systems –the chillers, switchgear, generators and cooling plant needed to keep data racks from overheating. According to research cited in the report, MEP equipment alone can account for up to 90 per cent of a data centre’s cumulative whole-of-life embodied carbon. Structural elements – reinforced concrete, steel framing, the substructure needed to bear the weight of heavy IT equipment – come a close second. Another area of consideration, although treated separately in life cycle analysis due to varied contractual arrangements and ownership models, is the computational infrastructure itself, particularly as AI chip manufacturing may cost more carbon due to rare earths, high-purity silicon and energy-intensive lithography.
Far more knowledge and capability exist when it comes to decarbonising structural components to data centres than MEP and computational infrastructure aspects. A lack of product specific Environmental Product Declarations (EPDs) and verified carbon data, made worse by bespoke project configurations and supply chains that outpace the speed at which third-party verification can keep up, makes accounting for and realising lower embodied carbon outcomes in MEP and computational infrastructure very tricky.
However, the pivotal time for data centres and decarbonisation is now. Australia’s mandatory climate-related financial disclosure regime, in effect since January 2025 under the Australian Sustainability Reporting Standards (ASRS 2), now compels large developers and operators to report Scope 3 emissions. It’s a structural shift that will force clearer accountability across MEP suppliers, structural contractors and IT hardware vendors alike.
Encouragingly, parts of the Australian industry are already moving. AirTrunk’s “Embodied Carbon Ratio” –tonnes of carbon dioxide equivalent per kilowatt of IT capacity delivered – has set a portfolio-wide threshold designed to embed embodied carbon considerations into design decisions before procurement begins. It’s an example of the kind of intensity-based metric MECLA argues the whole industry needs to standardise around, rather than continuing to rely on inconsistent, project-by-project assessments.
But embodied carbon is still missing from a lot of the national conversation when it comes to sustainable data centres. In March 2026, the Australian Government released its national expectations for data centres and AI infrastructure developers. Meeting them is, according to the government, the foundation of social licence to operate in Australia. But the expectations fail to mention embodied carbon alongside energy, water, the workforce and innovation.
Australia has the credentials and investment appetite to lead on a world class and best practice sustainable data centre delivery that is energy and water positive, in the most appropriate locations, contributes to local community development, and addresses its embodied carbon footprint. And the boom in data centre development in the nation, and the immense financing that follows it, offer a window of opportunity to drive upstream sectors in concrete, steel, MEP, IT infrastructure, and copper (among others) to decarbonise themselves as they service this rapidly growing demand.
Whether we capture it is now up to us.
