Australia is currently experiencing a data centre mania in which tech companies and developers feverishly rush to build hyperscale data centres across the country. Like mushrooms after rain, these expansive, anonymous buildings are sprouting up across city suburbs as well as regional centres with a pipeline of much greater scale in process. One estimate suggests $150 billion of investment in this sector by 2030.
Yet, just as this data centre frenzy hits top gear, cracks are appearing. As has occurred in the United States, local communities are rising up against the buildout. While data centre developers have enjoyed the support of state governments desperate for business investment and fast-tracking their projects as “state significant” and “critical infrastructure”, local communities have waged a David versus Goliath struggle to stop them.
In our research we have focused on four local communities fighting data centre proposals and have examined the issues of concern and also their campaign strategies and tactics.
In two of these cases, at Hazelmere in Western Australia and Lane Cove in NSW, community campaigns succeeded in having the developers withdraw their projects. In our third case in West Footscray in Victoria, community opposition to the extension of the NextDC data centre and blockades against the construction of 22 metre power poles, recently resulted in that company losing over 40 per cent of its share value. [On Monday the company said its chief executive Craig Scroggie would lose millions of dollars in potential future bonuses, and chairman Douglas Flynn will step down as a result of the write down – ed.]
Far from being just NIMBY concerns, these communities are rightly trying to protect their neighbourhoods from becoming heavy industrial heartlands whilst planning regulations fail to keep pace with technological change.
In our fourth case, local communities in the Southern Highlands are facing a proposal to build one of the world”s largest reciprocating gas power plants to fuel a giant data centre complex in the green pastures of Moss Vale, gaining significant media coverage and representations to the Federal Senate Inquiry into AI and Data Centres.
Each of these communities confront development proposals that impact directly on their quality of life, including the scale and proximity of facilities, noise and pollution, and declining property values.
Far from being just NIMBY concerns, these communities are rightly trying to protect their neighbourhoods from becoming heavy industrial heartlands whilst planning regulations fail to keep pace with technological change.
Concerns are diverse and cross partisan. Community members fear the cumulative impacts of unfiltered diesel emissions from data centre clusters on their families” health, the fire risk of millions of litres of diesel stored alongside banks of lithium-ion batteries, the potential for heat islands impacting flora and fauna, as well as risks of pollution to local waterways from PFAS and other chemicals.
Community groups are acutely aware that these facilities displace alternative more desirable land uses and offer minimal ongoing employment opportunities. And, of course, residents dread living with permanent exposure to the data centre’s interminable humming noise let alone the ongoing construction disruption as these centres are built and then extended in clusters.
But there are bigger questions
However, while it has been left to under-resourced local communities to challenge the claims of data centre developers and their government supporters, bigger questions are yet to be addressed.
Our research revealed a general acceptance of the digital infrastructure rollout, which is problematic, as it forecloses an important discussion about how we choose to allocate our finite resources, who gets to decide, and who will benefit (or lose out).
Are we directing resources away from more essential needs, such as building housing? There is no critical assessment of which uses of generative AI justify the diminution of water, energy, land and minerals.
When is a workload truly so important that it requires layers of redundancy to run 24/7 without skipping a beat? Could a different or simpler tool be used for the task? Is there a public benefit? How can we ensure we have transparent and fair reporting on resource use at the site level?
These are decisions that need to be made now, before the infrastructure is designed and built, and generative AI is embedded into every system we use by default.
A lack of transparency and disclosure by BigTech around the very material impacts of our digital lives is preventing the public scrutiny that could drive corporate accountability. Sustainable AI use requires us to interrogate which tasks should be curbed and how we can prioritise low-impact tools for everyday use. This isn’t an outright rejection of technology, but recognition of the need for responsible demand management and resource awareness in decision making across the full lifecycle from manufacturing, to use, to end-of-life.
And there’s climate change
Of course, an even more critical issue is what the embrace of genAI, and the data centres that drive it, mean for climate change. This week the UNEP confirmed the world is now on track for 2.8°C warming by the end of the century, leaving the Paris Agreement’s 1.5°C target in the rear-view window.
Dubious claims have been made by US tech giants about how AI would “solve climate change”, yet, what was surprising in our research was the lack of engagement with this critical issue. Local communities were busy enough fighting the local impacts of data centre development.
However, politicians and even clean energy advocates appeared nonchalant about the climate implications of an entire new industry consuming massive amounts of additional energy (estimated to be equivalent to the electricity consumption of Japan by 2030). For renewable energy advocates, the data centre boom offers the possibility of financial certainty for new wind and solar projects; for politicians the investment helps beleaguered state economies.
But questions over the emissions impacts of this industry”s energy consumption were often missing from debate. Data centre reliance on the electricity grid will place even greater pressure on existing generation, likely lead to further extensions in the life of ageing coal-fired power stations, and provide an incentive for the construction of new gas-fired generation, as is already being proposed in the Northern Territory and Queensland. And that is before we even consider how AI data centres enable even greater fossil fuel extraction.
Our research has highlighted that strategically-minded and hard-working community volunteers can successfully challenge data centres in their neighbourhoods and undermine the social licence of this new industry. However, this should not be their role. The risk is that data centre development will just be pushed into those communities that lack the capacity to engage in this fight; the “sacrifice zones” of the digital era. In the absence of adequate regulation, these developments also have a nasty habit of coming back meaning communities can’t afford to relax.
Civic society and non government organisations will need to heighten pressure on governments to regulate in the public interest.
This is not the time for pragmatism: the economic costs of the climate crisis will far outweigh any economic benefit we might squeeze from data centres. It’s clear that we need government to introduce infrastructure siting rules to protect communities. But it’s far more important we ensure that no data centre development can result in delays to climate goals and that promised legislation includes full and transparent disclosure of data centre operations and workloads.
How can an industry expect to have social licence if the public has no idea what is taking place behind those big, ugly walls?
