Submission – Getting it right: Building AI infrastructure that works for Australia

The Australian research and development (R&D) sector require a strong policy framework to ensure it is prepared for AI. This framework is needed to guide the development of the necessary infrastructure, ethical guidelines, and educational programs for Australian researchers to harness AI’s potential responsibly and effectively. A coordinated national approach to AI governance will provide clarity and consistency for researchers, support responsible adoption of AI technologies, and strengthen trust in AI-enabled research.

The consultation paper identifies the right risk – that Australia could host AI infrastructure and training while the research outputs, intellectual property and commercial returns accrue offshore. This submission focuses on three critical areas needed to address that risk: high-performance computing infrastructure, mechanisms to support the research and innovation system, and opportunities to strengthen skills, training and education.

The Academy’s central point is that hosting data centres onshore does not bolster AI capability on its own. Capability depends on Australia investing in the high-performance computing and data (HPCD) infrastructure that underpins AI and data-intensive R&D by both the research sector and industry.

The Academy welcomes the consideration of mechanisms to support Australia’s research and innovation system. As Australia is increasingly seen as an attractive destination for foreign data centre investment, this should be leveraged to strengthen domestic research, innovation and industry capability.

The Academy recommends:

• The Australian Government should develop a 10-year national high-performance computing strategy and investment plan to secure computing capability for AI and data-intensive research. The Academy proposes an investment of $200 million per year over 10 years to support sector planning, upgrade facilities, and coordinate investment in Tier-0 (regional) capability.
• Mechanisms to secure contributions to Australian research should include a levy on data centres to ensure this extractive industry builds local capability, and to provide merit-based access to compute capacity under favourable terms for all Australian researchers.
• Hyperscalers and frontier AI developers must contribute to the development of a skilled workforce in Australia. They should also be required to make commitments to uplift researcher AI literacy through education initiatives or funding for programs.
• Mandatory assurance and incident-reporting requirements should be implemented for high-risk AI-agent training and evaluation conducted in, or using infrastructure located in, Australia.
• Public reporting on data centre energy use, water use and emissions should be frequent and standardised, and published in open, machine-readable formats consistent with FAIR data principles.