Sharon AI Holdings (NASDAQ: SHAZ) has signed a six-year strategic AI infrastructure compute collaboration with NVIDIA, locking in a Master Cloud Services Agreement worth up to US$4.88 billion and a deployment of 40,000 Grace Blackwell GB300 GPUs across 72 megawatts of new Australian data center capacity.
72MW Of NVIDIA DSX Capacity In Australia
The collaboration centres on NVIDIA's reference DSX AI factory design, with Sharon AI building out liquid-cooled halls scaled to land the 40,000-GPU fleet. The Australian capacity is targeted at the country's AI startups, large enterprises and university research groups, where GPU supply has remained tight despite the wave of new local AI factory announcements.
Revenue-Share Economics And Credit Support
The deal is structured so Sharon AI can commit to large-scale NVIDIA infrastructure while sharing the upside: Sharon AI sells NVIDIA-powered cloud services under its own brand, NVIDIA earns standard product revenue, and the chipmaker takes a share of the cloud revenue running on the supported capacity. NVIDIA is also extending credit support that lets Sharon AI scale faster than its balance sheet alone would allow.
Pipeline Hits 132MW Of Contracted AI Capacity
Following the agreement, Sharon AI's total AI factory capacity expands to 132MW, of which 102MW is now contracted to end customers. The company expects more than 55,000 NVIDIA GPUs deployed by mid-2027 — a footprint that would put it among the larger sovereign-aligned GPU clouds outside the U.S. hyperscalers and rivals platforms such as OQC's London Genesis quantum-AI data centre and Samsung Foundry's 10th-gen Google TPU Icefish project.
The deal underscores how NVIDIA is increasingly anchoring sovereign GPU build-outs by combining hardware orders with cloud revenue-share — a model it has also used in partnerships referenced in coverage of OpenAI's expanded enterprise distribution.
Reporting based on coverage from BusinessWire, StockTitan, Converge Digest and Reuters.