Sharon AI has announced a landmark six-year strategic collaboration with Nvidia, valued at up to $4.88 billion, to build a massive "AI factory" in Australia.
At its core, this deal involves constructing a 72-megawatt data center designed to house up to 40,000 of Nvidia's state-of-the-art GB300 GPUs. This isn't just any data center; it's an 'AI Factory,' a standardized, high-performance computing campus based on Nvidia's DSX blueprint. The agreement includes a revenue-sharing model, where Nvidia reserves capacity that can also be sold to other clients, creating a flexible partnership.
This major commitment didn't happen overnight and was built on several key steps. First, Sharon AI secured a foundational $950 million contract with a major tech company in the Asia-Pacific region. This provided a crucial anchor customer to justify such a large-scale build-out. Second, the company raised $350 million through convertible notes, securing the necessary capital to begin purchasing the expensive GPUs and networking equipment. Finally, appointing a former CEO of Telstra as its chairman bolstered its governance, giving partners like Nvidia the confidence to engage in a long-term, high-stakes project.
The choice of Australia is also highly strategic. The Australian government is actively promoting a 'Sovereign AI' policy, which emphasizes developing local AI capabilities for security and economic independence. Sharon AI's new facility directly supports this national goal, positioning it to become a key provider for government agencies and regulated industries that require compliant, onshore AI infrastructure. This policy provides a powerful tailwind for the project's success.
However, the path forward carries significant challenges. The agreement comes with strict service level agreements (SLAs) for uptime and repairs, meaning any failure could lead to penalties or even termination. For a company still in its scaling phase, this execution risk is substantial. Furthermore, the project's large power requirement—72MW—plugs directly into Australia's ongoing debate about grid stability, and any delays in securing reliable power could impact the entire timeline and budget.
In essence, this partnership with Nvidia provides Sharon AI with a guaranteed supply of cutting-edge technology and a clear roadmap for growth. Yet, it also places immense pressure on the company to deliver flawlessly on one of the largest AI infrastructure projects in the region.
- AI Factory: A term for a standardized, end-to-end data center optimized for AI workloads, from the GPU clusters to the networking and software.
- Sovereign AI: A national strategy to build and control a country's own AI infrastructure and capabilities to ensure data security, economic competitiveness, and independence.
- GB300 GPU: Nvidia's next-generation GPU (Graphics Processing Unit), part of the Blackwell architecture, designed for massive-scale AI training and inference.