Australia’s ambition to be a global leader in artificial intelligence (AI) is clear—from sovereign AI strategies to booming enterprise deployment. But underneath the promise lies a pressing energy challenge. Data centres already account for approximately 5% of Australia’s electricity supply today, and according to reports could consume as much as 8% by 2030, or even escalate towards 15% under high-growth scenarios.

That shift isn’t simply a statistic, it’s a ticking time bomb for the grid, for emissions commitments, and for sustainable growth.

A Grid Under Pressure, a Landscape in Transition

The Australian Energy Market Operator (AEMO) has recently highlighted that overall grid demand has now surpassed pre-2008 peaks. Energy experts warn that skyrocketing demand from AI-powered data centres may necessitate an additional 3.3–5 GW of capacity by 2030 – raising wholesale electricity prices by up to 70% in peak periods.

These trends collide with ambitious climate policy. The Climate Change Act 2022 commits Australia to a 43% reduction in greenhouse gas emissions by 2030 and net zero by 2050. Meanwhile, starting January 2025, major enterprises must legally include sustainability metrics in their annual reports under amended Corporations Act provisions: AI deployments must now balance innovation with accountability.

The Hidden Carbon in Connectivity

Most attention has focused on AI compute or model efficiency, but overlooking the network is a costly mistake. Every AI-generated insight, prompt, or image travels relentlessly across data centres, clouds, and edge nodes. That makes the network a critical carbon and cost hotspot – and a strategic lever often hidden in plain sight.

As The Australian recently reported in its coverage of sovereign AI, energy limitations may hamper domestic AI infrastructure unless we modernise our networks and align digital ambition with physical capacity.

How Smart Networks Enable Sustainable AI

Network-as-a-Service (NaaS) platforms offer a dynamic alternative to traditional, over-provisioned infrastructure. By allowing enterprises to scale connectivity according to real-time demand, this platformisation approach cuts idle capacity, enhances performance, and significantly trims energy use. The sustainability benefits are tangible: efficiency gains are measurable, verifiable, and directly reportable in ESG disclosures. In short, a faster, greener network doesn’t just improve AI, it justifies it under today’s policy environment.

Here are five NaaS strategies Australian enterprises should implement

1. Edge-First Architectures
Locating compute closer to data origination points reduces latency and the energy cost of moving data across long distances.

2. Right-Size AI Models
Not all tasks require heavyweight models. Smaller, efficient models can operate closer to the edge, reducing both compute and network energy demands.

3. Adopt NaaS platforms
Move away from static network provisioning and legacy systems. NaaS allows for responsive, usage-based scaling, avoiding energy waste during idle periods.

4. Leverage Renewable Energy Integration
Partner with data centre providers that source power from renewables or use efficient technologies such as liquid cooling systems. Deloitte notes that in 2025, 90% of new data centre projects target top-tier NABERS environmental ratings, with liquid cooling delivering up to 10% energy savings.

5. Support Regulatory Collaboration
Policymakers, including the Net Zero Economy Authority and AEMC, are advancing grid reform to accommodate rising AI-related demand. Engaging with these stakeholders aligns commercial AI investment with public infrastructure planning.

A Strategic Imperative, Not Just Infrastructure

Australia’s path to AI leadership is clear – but not inevitable. Without modernising the network layer, AI will be throttled by energy constraints and policy friction. A smarter network unlocks efficiency, sustainability, and compliance, all while underpinning performance.

As media coverage on sovereign AI highlights, success depends on aligning ambition with infrastructure reality.

To prepare for the AI age, the most intelligent upgrades are not just the algorithms, but also the network they run on.