The real AI race is shifting from algorithms to infrastructure. Multi‑gigawatt data centres, reliable power and secure subsea cables now shape who captures value and who carries risk. Australia is an attractive Asia‑Pacific data‑centre market, but grid constraints and fragmented governance threaten its ambition to be a trusted Indo‑Pacific AI hub rather than a peripheral server‑farm landlord.

Australia’s National AI Plan makes strong commitments to infrastructure and adoption, but leaves governance and liability unresolved. This uncertainty exposes SMEs to risk, complicates investment decisions, and weakens public trust. History shows that technologies such as electricity became dependable only when standards, liability, and oversight aligned. AI now requires the same institutional attention if it is to move from experimentation to reliable national infrastructure

Australia’s next wave of innovation will depend on how clearly policy separates generic data centres from high‑performance computing (HPC). Bundling them into one “digital infrastructure” bucket directs capital to cloud‑style capacity while underinvesting in leadership‑class HPC needed for frontier AI, climate modelling, defence and advanced industry. The piece argues for explicit “AI‑ready HPC” and “AI‑capable cloud” tracks in national strategy, funding and governance.

The productivity lag from AI reveals competing logics: automation that replaces labour to cut costs versus augmentation that expands capability to create value. For innovation ecosystems, this choice is existential. Strict safety regulations risk creating moats favouring large incumbents, while unrestricted deployment threatens research quality. The narrow path requires public interest AI infrastructure, risk-weighted governance, and incentives rewarding high-impact discovery