en.Wedoany.com Reported - SouthernCrossAI (SCX) has announced its integration into the Equinix Fabric AI ecosystem, planning to deploy sovereign AI inference nodes across Australia using SambaNova SN50 reconfigurable dataflow units. These nodes provide enterprises, government agencies, and developers with private connection access via Equinix Fabric, preventing data from traversing the public internet and strengthening data sovereignty control. SCX CEO David Keane stated that any organization connected to Equinix Fabric can access these sovereign inference nodes through private direct connections.
SCX currently operates its first AI inference node in the Equinix SY5 data center in Sydney, with plans to add new nodes in Melbourne and Brisbane, expected to go live in batches between 2026 and 2027. Chris Johnston, Interim Managing Director of Equinix Australia, welcomed SCX's addition to the ecosystem. This collaboration further expands the vendor landscape of Equinix Fabric AI, while providing SCX with direct access to Equinix's global base of enterprise and government customers, accelerating the market penetration of its sovereign AI services.
Regarding hardware selection, SCX utilizes the SambaNova SN50 chip as its core computing unit. The SN50 features architecture optimized for AI inference tasks. According to published benchmark comparisons, its performance in agent-based inference workloads surpasses that of Nvidia B200 GPUs, while operating with significantly lower power consumption. It can be deployed directly within existing air-cooled data center environments without requiring modifications to cooling infrastructure. The chip supports models with up to 10 trillion parameters and a context window of 10 million tokens, along with millisecond-scale model switching capabilities, making it suitable for sovereign cloud node scenarios with concurrent multi-tenant, multi-model inference.
SCX positions itself as a full-stack sovereign AI infrastructure provider. Its platform is compatible with both open-source and proprietary models, including self-developed large language models optimized for Australian English and local use cases. The company explicitly commits to not retaining customer data, not routing workloads overseas for processing, and not using customer data to train models, thereby building a compliance-based differentiation against international public cloud service providers. The first Sydney node is already carrying early testing workloads for certain government agencies and enterprises. The subsequent launch of nodes in Brisbane and Melbourne will form an Equinix Fabric interconnected inference network across Australia's east coast, delivering access latency of under 5 milliseconds for local users.
Demand for sovereign AI is moving from proof-of-concept to scaled deployment, and SCX's deployment model offers a reference for localized computing infrastructure construction in overseas markets. By integrating dedicated AI chips, private interconnection networks, and a full-stack software platform within Equinix's neutral data centers, SCX seeks to carve out an intermediate market for sovereign inference between public cloud and private deployment, catering to industries with stringent data jurisdiction requirements, such as defense, financial regulation, and healthcare.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
