Sovereign AI needs solutions for the energy bottleneck
Sovereign AI is not only a matter of technological capability but also of national security and long-term economic resilience. It requires more than just models and algorithms.
Artificial intelligence is fast becoming the backbone of innovation across Australia, powering everything from public services to national defence capabilities. But behind this lies a critical dependency: infrastructure. Without sovereign control of the data centres, chips, and power systems that fuels our AI, Australia risks falling behind in the global digital race.
As AI’s influence grows, so does the need for Australia to develop and control AI technologies within its borders. That is, national AI governance and development. A concept known as ‘sovereign AI’.
Sovereign AI ensures systems reflect Australian values and ethics, comply with regulations, and support the country’s strategic interests.
Relying on overseas infrastructure to support our AI capabilities has the potential to impact our ability to retain economic value and independent control over critical digital infrastructure, including safeguarding our data.
In this way, sovereign AI is not only a matter of technological capability but also of national security and long-term economic resilience. But it requires more than just models and algorithms. It also needs chips, onshore data centres and electricity. Lots of electricity.
Data centres need 24/7, reliable, high-quality electricity supply with built-in redundancy. Deloitte’s TMT Predictions 2025 highlights that the rapid growth of AI data centres is already pushing operators to adopt more sustainable and forward-looking technology and energy solutions.
Elsewhere, some estimates have forecast that data centres could consume up to 15 per cent of the country’s electricity by 2030 under a high-growth projection. Even the more conservative projection estimates an 8 per cent share (up from 5 per cent today).
For Australia, this presents a real challenge. Building sovereign AI capability in any serious sense will require investment in the technology and energy infrastructure to power our data centres. If not, we risk importing capability through hyperscalers (large cloud service providers), ceding control and sovereignty in the process. Or alternatively (and possibly even worse) building data centres on a grid that is not equipped to support its needs.
At the heart of this challenge is a shift in the software-hardware dynamic. Historically, hardware has always led while software followed. Now, the coin has flipped, and AI development is surging ahead of infrastructure. Software capabilities are outpacing the physical systems required to support them. Hardware, energy, and networks are playing catch-up to software’s rapid evolution.
So, what technologies can we adopt to prevent this energy bottleneck constraining our nation’s sovereign AI capacity? One of the most promising pathways to manage the surging power demands of generative AI is improving chip-level energy efficiency. A new generation of gen AI–specific chips can now train advanced models in 90 days while consuming just 8.6 GWh, less than one-tenth the energy of prior-generation chips for the same task.
Both the private sector and government need to work together to secure a pipeline of chips from global manufacturers to ensure Australia can access the latest high-efficiency semiconductors.
Another key technology driving lower data centre energy usage is liquid cooling. It can reduce power consumption by up to 90 per cent compared to traditional air-based systems and is better suited to manage the intense heat generated by densely packed, high-performance AI chips. However, it also introduces water usage concerns, as AI data centres may require vast quantities of freshwater for cooling, a resource that is both finite and increasingly under pressure.
Balancing energy efficiency with sustainable energy production and responsible water use will be critical as this technology scales. In Australia, one solution is to co-locate data centres with renewable energy infrastructure. Projects like Snowy Hydro 2.0 provide access to water, energy and grid infrastructure that can grow alongside data centre demand. Similarly, positioning centres near wind or solar farms offers an opportunity to use clean power while reducing the need for additional infrastructure.
Offloading AI workloads to edge devices is another tool to manage power demands. This is especially effective for applications where latency is crucial or where sensitive data and privacy needs are high.
By processing data locally, edge computing reduces reliance on central data centres and limits the transmission of sensitive information across networks. This not only conserves energy and reduces network strain, but also strengthens data security by keeping information closer to its source. As edge capabilities grow, this distributed approach will enable a more efficient and secure balance between edge and core infrastructure.
Australia must align its digital ambitions with its physical infrastructure capacity. As the nation advances towards sovereign AI capabilities, energy availability and management represent significant constraints. Through deliberate frameworks and robust public-private collaboration to drive strategic infrastructure investment and technologies, Australia can establish the foundation necessary to support a secure and sustainable AI ecosystem.
Peter Corbett is National Telecommunications, Media and Technology Industry Lead Partner at Deloitte Australia.
-
Disclaimer
This publication contains general information only and Deloitte is not, by means of this publication, rendering accounting, business, financial, investment, legal, tax, or other professional advice or services. This publication is not a substitute for such professional advice or services, nor should it be used as a basis for any decision or action that may affect your business. Before making any decision or taking any action that may affect your business, you should consult a qualified professional adviser.
Deloitte shall not be responsible for any loss sustained by any person who relies on this publication.
About Deloitte
Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee (“DTTL”), its network of member firms, and their related entities. DTTL and each of its member firms are legally separate and independent entities. Please see www.deloitte.com/au to learn more.
Copyright © 2025 Deloitte Development LLC. All rights reserved.
-
