TL;DR

• SCX.ai is delivering the world's first RDU-based sovereign AI cloud to Australia, built with SambaNova.
• The platform is hosted onshore in Australia, and data never crosses national borders.
• It runs on SambaRack SN40-16, powered by SambaNova's fourth-generation Reconfigurable Dataflow Unit (RDU), at an average of 10 kW per rack.
• SCX.ai offers inference-as-a-service (IaaS) to enterprises and government agencies across Oceania.
• It serves current open-source models, including OpenAI's gpt-oss and DeepSeek.

Background

AI sovereignty became a procurement requirement across Oceania. Australian enterprises and government agencies now need to know where their inference runs, which law governs it, and who could compel access to it. For too long, running frontier AI meant compromising on data control by depending on foreign cloud providers.

SCX.ai set out to change that: A sovereign AI backbone hosted onshore, built on the world's most efficient AI hardware, delivering inference as a service without sending data offshore.

Challenge: A Sovereign AI Backbone, Hosted Onshore

SCX.ai's build had to satisfy several requirements at once.

• Data residency that survives audit. Inference has to happen onshore in Australia, with data that never crosses national borders.
• Efficiency at the hardware level. A sovereign backbone for a national market has to run on hardware that is efficient enough to make the unit economics work.
• Inference-as-a-service. Enterprises and government agencies needed managed access rather than infrastructure to build themselves.

Solution: SambaRack SN40-16 and the Reconfigurable Dataflow Unit

SCX.ai standardized on the SambaRack SN40-16, built on SambaNova's fourth-generation RDU.

World's First RDU-Based Sovereign AI Cloud

SCX.ai is delivering the world's first RDU-based sovereign AI cloud to Australia. It is hosted onshore, and data never crosses national borders, so compliance with national values is built into where the platform runs rather than configured after the fact.

10 kW Per Rack

A SambaRack SN40-16 runs large models at an average of 10 kW, on hardware SCX.ai describes as the world's most efficient AI hardware. That efficiency is what makes a national sovereign backbone viable.

Inference as a Service for Oceania

SCX.ai offers IaaS to enterprises and government agencies across Oceania, serving current open-source models including OpenAI's gpt-oss and DeepSeek. The operator absorbs the compliance and infrastructure work rather than passing it downstream.

Why SCX.ai Chose SambaNova

1. Efficiency. An average of 10 kW per rack on hardware SCX.ai calls the world's most efficient AI hardware.
2. Jurisdictional control. Hosted onshore in Australia, with data that never crosses national borders.
3. World-first architecture. The world's first RDU-based sovereign AI cloud.
4. Open-source breadth. Current open-weight models, including gpt-oss and DeepSeek.

What Comes Next

SCX.ai is taking waitlist registrations ahead of service opening. The pattern is set: a sovereign AI backbone hosted onshore, compliant with national values, and built on efficient hardware. Read more on sovereign AI and national autonomy, or talk to our team about a sovereign deployment in your region.