Eight units arrive
The first wave has landed.
Today, our initial shipment of SambaNova SN40 units flew into Scotland — 8 systems forming the first tranche of a 150-unit deployment scheduled for this year. Delivered rapidly and efficiently with the support of SambaNova, this marks a significant step forward in bringing scalable AI infrastructure online in the UK.
But these aren’t just another set of AI boxes. SambaNova systems are purpose-built for large-scale AI inference and model serving — delivering high performance without the traditional infrastructure burden. They are:
Low power compared to GPU-heavy alternatives
Air-cooled — no requirement for complex water cooling
Deployable in standard data centre environments
In short, they bring advanced AI capability out of specialist facilities and into mainstream infrastructure.
Over the coming days, these units will be transported to a major UK colocation data centre, where we will:
Install, configure, and commission the systems
Integrate high-speed connectivity
Establish direct linkage to AWS and GCP environments
This hybrid architecture combines dedicated AI infrastructure with hyperscale cloud flexibility — delivering both performance and reach.
Initial services will launch with:
DeepSeek bundle
MiniMax bundle
Over the coming months, we will introduce a sovereign AI service powered by these initial units — enabling secure, locally governed model deployment and inference, with capacity expanding on demand as additional SambaNova units from our 150-system order arrive in the UK on a rolling ~120-day delivery cycle from initial order to full deployment.
Developers and partners exploring these systems are invited to go further:
Sample additional models via SambaNova Cloud
Import and run their own models on our infrastructure
Transition workloads into a sovereign environment as required
Alongside this, we will be demonstrating additional products and capabilities as the platform comes online — giving us the flexibility to work closely with clients who have specific technical, regulatory, or performance requirements.
These first 8 units are the foundation layer — the opening move in a distributed deployment model. We will place capacity across major colocation facilities for clients already resident, while also supporting dedicated standalone deployments for organisations requiring their own sovereign infrastructure.
From arrival to activation, the platform is coming to life.
