Positioning Australia's Data Centre Strategy for the AI Era. Latency, Energy, Water, and Policy Considerations for Sustainable Hyperscale Growth

Artificial intelligence is driving an unprecedented build-out of data centre infrastructure, and national hosting strategies now shape energy transitions, water security, and digital sovereignty. Australia hosts one of the world's largest data centre markets relative to its population, yet its positioning remains contested between serving domestic demand, anchoring the Indo-Pacific digital economy, and hosting latency-tolerant global AI workloads. This paper assesses how Australia should position its data centre sector for long-term hyperscaler requirements using open data and reproducible analysis. A physical latency model built on great-circle geometry, fibre propagation, and an open submarine cable dataset quantifies which classes of AI workload Australia can serve at global, regional, and domestic scales. Scenario analysis calibrated to Australian Energy Market Operator reporting projects electricity demand of roughly 10 to 13 TWh per year by 2030, rising to as much as 12 per cent of 2024 National Electricity Market consumption by 2035 in the high scenario, with water consumption in the tens of gigalitres depending on cooling efficiency. Every derived quantity was independently recomputed and validated. The results show that physics confines Australia's real-time AI role to its own territory while supporting a strong position in model training, sovereign inference, and interactive services for the Indo-Pacific, and that energy and water constraints prove manageable only under deliberate cross-portfolio coordination. Situated against the national Expectations framework of March 2026 and the Australian AI Standards announced in July 2026, the paper derives recommendations for siting, grid integration, water stewardship, community protection, and edge investment relevant to resilient and sustainable infrastructure planning in the built environment.