SambaNova Systems, a U.S.-based provider of next-generation AI infrastructure, has announced three landmark agreements with partners across Australia, Europe, and the United Kingdom to power sovereign AI inference clouds that promise to redefine how nations deploy high-performance artificial intelligence within secure, sustainable, and locally governed environments.
The partnerships — with SCX in Australia, Infercom in Europe, and Argyll in the UK — mark a significant step in establishing regional AI ecosystems designed to meet stringent data sovereignty, compliance, and energy-efficiency goals. Each deployment will run on SambaNova’s Reconfigurable Dataflow Unit (RDU)-powered SN40L systems, which the company claims deliver unprecedented inference performance at just a fraction of the energy consumption of conventional GPU-based systems.
“These partnerships across three continents represent a fundamental shift in how the world approaches AI sovereignty, sustainability, and security,” said Rodrigo Liang, CEO and co-founder of SambaNova. “In a year when AI has pushed both policy and power grids to their limits, SambaNova is delivering a new kind of AI cloud — one that’s private, governed locally, and radically more energy-efficient.”
The timing of these deals is crucial. AI inference — the process of running trained AI models to deliver outputs — is rapidly surpassing training as the dominant workload in the global AI landscape. As inference demand grows, so does the strain on power grids and infrastructure. SambaNova’s approach directly addresses this challenge, offering an architecture capable of delivering high-throughput inference while dramatically reducing energy draw.
In Australia, SCX will launch the nation’s first ASIC-based sovereign AI cloud, powered entirely by renewable energy and built in collaboration with SambaNova. According to SCX CEO David Keane, the partnership allows Australian organizations to achieve world-class AI performance without sacrificing data sovereignty or sustainability. “Traditional GPU systems can draw up to 120kW per rack, straining infrastructure and budgets,” said Keane. “In contrast, SambaNova’s SN40L systems consume just 10kW per rack — eliminating the need for liquid cooling, power upgrades, and drastically cutting operational costs.”
In the United Kingdom, Argyll will deploy the country’s first containerized sovereign AI inference cloud, powered by wind, solar, and wave energy. The infrastructure aligns with the UK’s national AI and data sovereignty strategy, ensuring that British enterprises and government agencies can innovate responsibly within domestic borders. “By working together with SambaNova, we’re delivering a pioneering inference cloud powered entirely by renewable energy,” said Peter Griffiths, Executive Chairman of Argyll. “It meets the highest standards of data protection, compliance, and operational efficiency — all within the UK’s own energy and policy framework.”
Meanwhile, Infercom is set to launch Germany’s first sovereign AI inference platform, providing fully EU-compliant, GDPR-safe AI infrastructure for enterprises, startups, and public sector clients. “Trust is the foundation of the next wave of AI innovation,” said Infercom CEO Altug Eker. “Through our strategic partnership with SambaNova, we’re advancing sovereignty, compliance, and efficiency — ensuring our clients benefit from AI solutions that respect privacy, meet European standards, and drive sustainable growth.”
Industry experts have highlighted the importance of energy efficiency as AI continues its rapid expansion. “You could say that AI infrastructure is in its incandescent phase; meanwhile companies like SambaNova have just invented LED technology,” said Peter Rutten, Vice President of Performance Intensive Computing at IDC. “It’s only a matter of time before the market recognizes the enormous cost and energy savings this architecture delivers.”
SambaNova’s RDU chip enables high-throughput inference at a fraction of the energy cost of legacy GPUs, allowing nations and enterprises to scale AI responsibly without overhauling existing data centers or compromising on sovereignty.
