SCX.ai & DDN expand Australia's sovereign AI cloud

SCX.ai and DDN have partnered to expand Australia's sovereign AI inferencing cloud, combining SCX.ai's domestic infrastructure with DDN's data platform.

The partnership targets Australian enterprises, government agencies and research institutions that want AI workloads and data to remain in Australia. The combined system is intended to support multi-tenant AI use while meeting rising demand for locally based infrastructure.

The announcement comes shortly after SCX.ai's debut on the Australian Securities Exchange following a fully underwritten AUD $40 million initial public offering. It described itself as Australia's first ASX-listed pure-play sovereign AI inference infrastructure business.

As it enters public markets, SCX.ai also disclosed recent commercial metrics. Contracted annual recurring revenue stood at AUD $6.5 million at the end of July, up 20.9% since May, while the platform had more than 400 active users.

Its first sovereign AI node is already operating at the Equinix SY5 data centre in Sydney. Adding DDN's Infinia platform is intended to strengthen the data layer for heavy concurrency and high-throughput inferencing.

According to the companies, the system is designed to reduce data input and output delays that can limit AI performance. They said the platform delivers sub-millisecond latency and can load key-value cache data up to 27 times faster, helping customers run large context windows and agentic AI workloads within Australia.

Power use

The announcement also highlights a broader constraint facing AI infrastructure projects: energy and cooling requirements. SCX.ai said its infrastructure uses SambaNova processors and is designed to generate more AI compute from the power and floor space available in existing commercial data centres.

It also said its architecture relies on air cooling rather than the water-intensive systems used in many conventional AI data centre designs. Ahead of its stock market listing, SCX.ai disclosed testing that showed its SambaNova-based infrastructure delivered about 2.5 to 5.6 times the performance per watt of GPU-based systems across selected stable inference workloads, depending on the model and hardware configuration tested.

That focus on efficiency is central to SCX.ai's case for building sovereign AI infrastructure in Australia rather than relying solely on offshore cloud providers. Domestic deployment has become a growing priority for organisations concerned with data control, privacy obligations and local operational oversight.

David Keane, Founder and Chief Executive Officer of SCX.ai, described the partnership as a step towards a fully domestic AI cloud for Australian users.

"For the first time, Australian organisations have access to a fully domestic, enterprise-grade AI cloud that scales effortlessly," said David Keane, Founder and Chief Executive Officer of SCX.ai. "By integrating DDN's world-class data platforms into our architecture, we are ensuring that our infrastructure can handle the most data-intensive workloads on the planet. This partnership supercharges our mission to deliver advanced AI entirely onshore, without compromising speed, security or sovereignty."

Data layer

DDN framed the deal around the role of storage and data movement in AI system performance. As inferencing workloads grow, the speed at which systems can access and move data has become a constraint alongside processor availability.

Sven Oehme, Chief Technology Officer at DDN, said the partnership is intended to address that bottleneck.

"Sovereign AI isn't just about where data resides, it's about whether a nation can operate AI at scale, efficiently and without compromise," said Sven Oehme, Chief Technology Officer at DDN. "Our partnership with SCX brings together purpose-built inferencing infrastructure with DDN's Infinia data platform to remove the single biggest barrier to AI performance: the data layer. By eliminating I/O bottlenecks and enabling extreme concurrency with sub-millisecond latency, we're fundamentally changing the economics of inference - delivering higher utilisation, lower cost per token and true enterprise-grade sovereignty. This is how countries turn AI ambition into real, operational capability."

The increased cloud capacity is also expected to support SCX.ai's own model development, including Project MAGPiE, which it describes as a sovereign large language model tuned for Australian cultural and commercial use. Alongside custom enterprise deployments, the company is preparing further node expansion as it builds a broader national network.

A second node is now being deployed following the launch of the Sydney site. SCX.ai said its first node at Equinix SY5 is operational and that it is progressing further locations as part of a sovereign AI inference network.