NVIDIA Tests Recurring AI Data Center Model With Australian GPU Expansion - Yahoo Finance

NVIDIA has initiated a new capital-efficient AI data center partnership in Australia, deploying 72MW of GPU infrastructure through a collaboration with SharonAI. This model shifts away from traditiona... Read original →

NVIDIA’s usage-based AI data center rollout in Australia signals a strategic pivot from chip vendor to infrastructure-as-a-platform operator. Technically, this accelerates co-deployment of 3nm EUV GPUs with liquid cooling and optical interconnects, forcing local data centers to upgrade power and thermal standards. While Australia poses low export control risk today, such asset-light models could become geopolitical hedges if U.S. AI chip restrictions expand geographically. Competitors like AMD and Intel will likely fast-track revenue-sharing cloud deals for MI300 and Gaudi3, while hyperscalers may restrict external access to custom AI chips to contain NVIDIA’s ecosystem reach. If validated in Australia over the next 12–24 months, this model will proliferate across Southeast Asia and the Middle East, shifting global AI capex from upfront CAPEX to recurring OPEX—and redefining semiconductor valuations around revenue predictability rather than peak compute specs.