Sharon AI Backs Remote AI Data-Center Model as Secured Capacity Reaches 212 MW
Sharon AI says its secured AI-factory capacity has increased to 212 MW as it argues that power-rich remote regions can support more AI workloads.
Why it matters: The story tests a major AI-infrastructure assumption: some compute may move toward regions where power and land are abundant rather than where end users are concentrated.
Australian AI-cloud provider Sharon AI is making the case that the next wave of artificial-intelligence infrastructure does not necessarily need to be concentrated around major cities. Chief executive James Manning said power-rich remote areas such as Western Australia's Pilbara could support AI workloads that do not require millisecond-level proximity to end users. The company says its secured AI-factory capacity has increased by 80 megawatts to 212 MW for deployment across 2026 and 2027. Second-quarter revenue increased 412% year over year to $1.9 million, while total contract value reached approximately $8.8 billion. Sharon's expansion is heavily tied to Nvidia infrastructure. Under a six-year collaboration disclosed in June, Sharon and Nvidia plan to enable 72 MW of Australian data-center capacity and deploy as many as 40,000 Grace Blackwell GB300 GPUs. The company has also signed several large multi-year cloud contracts. The emerging thesis is that advantage in AI infrastructure may increasingly depend on securing power, fiber and GPU capacity rather than only premium metropolitan data-center space. Sharon still faces substantial execution risk because its reported contract pipeline is many times larger than its present revenue base.