Balancing Bytes and Carbon: Australia’s Data Centre Dilemma

Abstract

Australia faces a strategic inflection point in its digital evolution, balancing the rapid expansion of hyperscale data centres with its climate commitments and sovereignty imperatives.

As AI and data-driven services surge across defence, healthcare and finance, the environmental toll of compute – electricity, water, emissions and e-waste threatens to undermine national targets. Simultaneously, data has become a sovereign asset, demanding jurisdictional control and operational discipline.

This paper argues that infrastructure innovation alone is insufficient. Instead, operational governance, embodied in Systems of Work like DOLIUM—is essential to reconcile sustainability with strategic ambition. DOLIUM enables organisations to eliminate redundant data, enforce retention policies and schedule compute responsibly.

It supports sovereignty by maintaining local control over sensitive datasets and complements infrastructure advances like liquid cooling and heat reuse.

Australia’s path forward lies in pairing sovereign infrastructure with auditable governance. By shifting from “store everything, compute anything” to “store what matters, compute what’s responsible,” the nation can scale digital capability without compromising climate goals or sovereign control. The success of this strategy will hinge not on the size of its data centres, but on the discipline of what they store and when they compute.

Discussion

It’s a pivotal moment in Australia’s digital evolution and our dependence upon cloud storage; as artificial intelligence and data-driven services accelerate across defence, healthcare, education and finance, the country is rapidly positioning itself as a regional powerhouse for hyperscale data centres.

Yet this growth comes with a paradox. Australia has committed to reducing greenhouse gas emissions by 62–70% by 2035, a cornerstone of its climate strategy for achieving the Paris Agreement target of net-zero by 2050. On the surface, building more data centres (energy-intensive facilities that demand constant power and water) risks undermining that commitment.

At the same time, data itself is becoming a strategic asset. From battlefield intelligence to genomic research, the power of data is reshaping national security, economic competitiveness and public service delivery. Ensuring that this data is stored, processed and protected within Australia’s jurisdiction is no longer optional—it’s essential.

Left unchecked, the growth of compute risks undermining emissions commitments and local utilities while creating expensive trade-offs between sovereignty and sustainability. The challenge is clear: how can Australia scale its digital infrastructure without compromising its environmental goals or sovereign control?

The Environmental Cost of Digital Growth

The digital boom comes with a carbon cost. Data centres are inherently energy-intensive, requiring constant power for servers, cooling systems and redundancy protocols. Even idle data consumes electricity. AI workloads amplify this demand, using 50-60% more energy than traditional computing. By 2035, data centres could consume 12% of Australia’s electricity, surpassing healthcare and placing immense pressure on a grid not designed for such high-load, continuous demand.

Water usage is another concern. Sydney’s data centres already consume 3.5 billion litres of drinking water annually, with forecasts suggesting this could rise to 25% of the city’s supply by 2035. Cooling systems, especially evaporative and open-loop designs, are major contributors and AI accelerates the need for thermal management as high-performance chips generate more heat.

Globally, data centres could emit 2.5 billion tonnes of CO₂-equivalent by 2030 . AI model training alone can produce emissions comparable to driving a petrol car 10-30km, per million tokens processed. Add to this the environmental toll of hardware production and e-waste (where a single 2kg computer requires 800kg of raw materials) and the urgency becomes clear.

Mitigation Starts with Smarter Data

If infrastructure is the supply side, governance is the demand control. The decisive interventions are therefore necessary to minimise unnecessary storage and processing, schedule heavy workloads intelligently and eliminate duplication and “shadow” datasets that quietly multiply infrastructure demand.

This is where platforms like DOLIUM offer a transformative solution. DOLIUM helps organisations identify and eliminate “shadow data”, redundant, obsolete or unused information that silently inflates storage needs and energy use. It maps data lineage, analyses usage patterns and enables defensible deletion based on business value and policy. Instead of blanket retention, organisations are empowered to implement targeted retention and tiering: only business-critical data consumes hot storage and active compute.

A Circular Vision for Infrastructure

Beyond energy savings, DOLIUM supports digital sovereignty by helping sectors like Defence, healthcare and education manage sensitive data without over-reliance on hyperscale providers. It aligns with climate goals by reducing emissions tied to unnecessary storage and it delivers cost efficiency by cutting infrastructure and operational overhead.

Combined with innovations in liquid immersion cooling, waste heat recovery and renewable colocation, platforms like DOLIUM offer a pathway that doesn’t sacrifice strategic ambition for environmental responsibility.

Sovereignty: The Paradox and the Practical Path

Australia’s Sovereign Data Centre Policy and the ASD Hosting Certification Framework reflect a clear intent: certain tiers of government data must live and be governed within Australia. But sovereignty isn’t only about rack location or vendor nationality, it is also about control over process and movement of data.

DOLIUM’s referential model provides a pragmatic middle path. Instead of wholesale replication of all on-prem data into hyperscaler ecosystems (with attendant carbon and sovereignty costs), organisations can maintain local control of canonical indices and only expose the minimal contextual artefacts external systems need to operate. This reduces data duplication, lowers energy use and retains legal control.

Sovereign cloud agreements (including large-scale partnerships with global providers) remain strategically useful for scale and specialised services. But pairing those infrastructure relationships with strong, auditable Systems of Work is what turns contractual sovereignty into operational sovereignty.

The Road Ahead

The right future requires both public policy and private practice:

For Government

• Make metadata-driven governance a procurement requirement for public AI projects (project approvals should include lifecycle tagging, retention rules and energy/water attribution).
• Tie sovereignty certification to demonstrable process governance: vendors should show not only where data sits, but how retention, provenance and scheduling are enforced.
• Support pilots that pair sovereign compute clusters with district heat reuse, prioritising locations where waste heat has nearby industrial demand.

For Enterprise and Research

• Implement System of Work patterns for data lifecycle, workload scheduling and telemetry before sprawling dataset migration projects begin.
• Use readiness scoring to prioritise automation pilots where environmental cost is low and governance maturity is high.
• Require dataset signing and provenance records for any external training or RAG index ingestion.

For the data-centre industry

• Expose APIs that provide renewable-share forecasts and water-impact signals so orchestration engines can make placement decisions in real time. Partner with SoW platforms to integrate environmental placement logic.

Closing – A dual path to sovereignty and sustainability

Australia does not need to choose between digital sovereignty and climate commitments. The right combination of infrastructure innovation (liquid cooling, heat reuse, renewables) and operational governance will let the country scale compute responsibly.

A System of Work like DOLIUM, is the strategic control plane that makes this possible. It shifts organisations from “store everything, compute anything” to store what matters, compute what’s responsible. By making retention and compute policy explicit, auditable and machine-enforceable, SoW reduces unnecessary energy and water use, enables smarter workload placement and secures sensitive data without needless duplication.

If Australia’s data-centre strategy succeeds, it will be because it disciplined what it stores and when it computes- not because it built the biggest warehouses of racks. DOLIUM-style Systems of Work are the governance tool that helps translate that discipline into operational reality.