By Seequent Vice President APAC, Karl Howley
While Australia’s new National AI Plan is a step towards the Government’s ambition to ‘build an AI-enabled economy’ that harnesses AI’s potential ‘for all Australians’, it may have missed a trick in not considering a significant risk to the growing AI data centre opportunity.
In today’s digital age, data centres are the backbone of our connected world, powering everything from cloud services to AI-driven applications. Hyperscalers such as Microsoft, Google and Amazon Web Services are building bigger and faster than ever, with McKinsey estimating $10 trillion will be spent on data centre infrastructure globally by 2030.
Australia has emerged as a data centre destination second only to the US, attracting $10 billion in investment in 2024 thanks to its stable operating environment, land availability, and proximity to growing economies. Announced in December, the National AI Plan seeks to maintain Australia’s leadership position amid the AI data centre boom.
Despite seeking to position ‘Australia as a leading destination for data centre investment while ensuring growth is sustainable and secure’, the National AI Plan does not consider subsurface risk in relation to data centre builds. While the opportunity to address subsurface risk in the AI plan has passed, it is not lost. The Government is developing a set of ‘data centre principles’, so construction is aligned with national interests. In my view, enabling effective subsurface risk management is squarely in the national interest.
The unseen threats to uptime
Before the first server is racked, a critical decision is made: where to build. Too often, this decision is based on surface-level factors such as land cost and fibre access, while the geological realities below are overlooked. Building on unstable ground is equivalent to constructing a house of cards. What happens when unforeseen ground conditions, for example encountering contaminated soil that needs to be remediated, lead to costly construction delays and increased material waste? Or worse, when subtle ground shifts or seismic tremors compromise the structural integrity of a facility designed for 100 per cent uptime?
Changing groundwater levels can impact the stability of foundations and even cripple infrastructure. A subsurface-informed digital twin enables site selection and operation with a deep understanding of these evolving strategic threats is foundational due diligence. Digital twins that incorporate subsurface understanding are a pathway to quantifying these risks, evaluating mitigation strategies and communicating this critical information to stakeholders.
Managing subsurface risk
The weight of hyperscale infrastructure (servers, cooling and power systems) demands deeper subsurface investigation than typical buildings. Considerations should include:
- A detailed desktop study to identify potential geohazards.
- A tailored ground investigation: borehole, cone penetrating radar, geophysics and lab testing.
- The development of an engineering ground model:
a) integrated geological + geotechnical interpretation
b) structure, lithology, groundwater, weathering profiles
c) geohazard mapping
d) 3D visualisation
e) geotechnical parameter assignment
The AI power paradox
As the AI revolution plays out, today’s data centres are getting larger and more complex. Increasingly they are designed using a digital twin to simulate in a digital space what is happening with compute load, the effect that it has on the utility requirements such as energy demand and its impact on the power grid. Data centres are already a significant energy user and Australia’s Energy Market Operator has projected data centre energy demand to triple by 2030, while experts warn national climate goals may be at risk.
Creating reliable renewable capacity available 24/7 is the smart solution, and the subsurface is part of the answer, for those willing to look.
Next-generation geothermal energy, pioneered by companies such as Fervo Energy, is a game-changer opening up new possibilities for countries like Australia. In a landmark project, Fervo successfully developed an EGS (Enhanced Geothermal System) that now delivers 24/7 carbon-free energy to Google’s data centres in Nevada. EGS borrows technology from the oil and gas sector to unlock geothermal energy in areas previously thought too costly or technically challenging. Fervo used Seequent’s advanced subsurface modelling to support drilling horizontal wells into previously inaccessible hot rock, creating a reliable, ‘always-on’ power source. Google is currently working with the University of Newcastle to reassess Australia’s geothermal potential in light of such EGS innovation.
From vulnerability to resilience
Considering the subsurface is not a matter of choice; it is fundamental to a data centre’s long-term viability. Every aspect of the lifecycle, from the stability of its foundations, and the sustainability of its power and water sources, depends on a deep understanding of the ground beneath.
But are these risks being taken seriously? Subsurface risk, environmental and renewable energy considerations must be in the mix if the National AI Plan is to achieve its goal of ‘capturing the opportunity’ by creating smart data centre infrastructure, as the demand for AI continues to climb. Incorporating subsurface risk considerations into the Government’s ‘data centre principles’ can help Australia to maintain its AI infrastructure edge and remain a global leader, as the fourth industrial revolution continues at pace.
Karl Howley is Vice President, APAC, for Seequent, a leading global developer of subsurface modelling software used on several Australian infrastructure projects.
