SambaNova is proud to support Argyll Data Development (ADD) launching the UK's first large-scale AI inference service powered entirely by renewable energy. This isn't just infrastructure; it's a sovereign AI revolution built on the SambaRack platform that delivers industry-leading performance at just 10 kW per rack while keeping Britain's data securely within its borders. We sat down with Peter Griffiths, founder and chairman of ADD, to discuss their plans for powering UK’s first sovereign sustainable AI inference service.
Who is Argyll?
Argyll is a Scotland-based data center operator with three years of experience in data development services. Our primary focus is delivering computing services across the UK with a strong commitment to green energy. Scotland’s abundant renewable resources — including wind farms, wave energy, and hydroelectric power — provide a strategic advantage for our operations.
Traditionally, data centers were built near demand centers, requiring minimal data cabling but extensive power infrastructure. With AI advancements driving higher power costs, we advocate for locating data centers near sustainable energy sources and extending data cabling to markets. Cloud technologies and improved data communications enable this approach, allowing ADD to power green data centers with minimal environmental impact.
What about SambaNova attracted Argyll to collaborate, and how does SambaNova fit into your vision?
The AI market has historically relied on high-power solutions like NVIDIA, which can consume up to 120 kW per rack. For large hyperscale developments, this translates to 600 MW to 800 MW of power. We chose SambaNova for its exceptional power efficiency, which aligns with our sustainability goals.
You note a market shift toward inferencing over training. Why the focus on inference?
Inferencing requires significantly less power than training, reducing costs for developers and avoiding energy waste. This efficiency allows us to deploy rapidly — in as little as six months — rather than undertaking multi-year builds. Speed is equally critical: Inferencing minimizes latency for time-sensitive sectors like healthcare and banking. As the UK’s inference market grows, its potential to disrupt traditional GPU-based AI approaches —through lower power consumption and faster transactions—makes it strategically vital.
Why is sovereign AI critical to UK businesses and governments?
Unlike traditional cloud computing, sovereign AI must reflect regional values and contexts. Applying a one-size-fits-all model (e.g., California or Australia) is ineffective for Scotland’s unique needs. AI-driven decision-making requires local embedding to address sector-specific requirements in life sciences, education, banking, and logistics. Our sovereign capability delivers proven, high-speed technology that lets organizations develop and deploy their own models—a frequent customer request.
Which vertical applications excite you most for the Argyll cloud?
We anticipate early adoption in life sciences (e.g., medicine) and banking due to their rapid AI integration. Logistics and customs management will follow, especially given post-Brexit paperwork demands. Education is another priority to train future AI talent. We also see transformative potential in agriculture: For example, our partnership with Intelligent Growth Solutions uses AI-controlled vertical farming to grow 4.5 million trees annually. Data center waste heat can even be repurposed here, creating a circular economy.
How are you deploying data centers with renewable energy?
We prioritize three approaches:
- Grid-connected sites: Leveraging Scotland’s wind-rich grid for deployments like SambaNova, which requires modest power, accelerates timelines.
- Schneider microgrids: AI-managed systems with private-wire renewable connections, powering data capacity or EV charging. One site is operational; another at Kilallan Farm (Dunoon) will support hyperscale needs like disaster recovery and banking.
- Wave energy: For Kilallan’s 600 MW requirement, we’re partnering with a Swedish firm to develop wave energy off Scotland’s west coast — aligning with Scotland’s nuclear-free policy while ensuring predictable output.
What are you excited about in the near future?
We’re focused on creating skilled jobs in Scotland. For instance, our partner Taylor & Fraser now trains 400 apprentices in electrical services and design. We also see AI transforming sectors like air traffic control simulation and gaming. By integrating AI into education, we’ll equip the UK workforce for long-term roles in this rapidly evolving field.
What use cases are you excited to deliver to the UK market?
Our team has extensive experience with Hume.AI, including a past air traffic control project where voice replication was critical. Hume.AI’s ability to interpret and respond through rule-based voice interactions is compelling. We plan to deploy it initially for training simulations, then scale to real-world applications.
Conclusion
SambaNova and Argyll Data Development are pioneering the UK’s first sovereign AI inference cloud — powered by 100% renewable Scottish energy and SambaNova’s ultra-efficient Reconfigurable Dataflow Units (RDUs) — to deliver high-performance inference at just 10 kW per rack. This partnership redefines scalable, sovereign AI for British enterprises and government agencies: Developers gain frictionless access to open-source models (like gpt-oss and DeepSeek) with near-zero latency, while all data remains securely within UK borders, ensuring compliance with UK GDPR and eliminating sovereignty risks. Launching Q1 2026, this isn’t just infrastructure; it’s a blueprint for sustainable, sovereign AI that prioritizes planetary responsibility alongside technological ambition.
SambaNova + Argyll = Sovereign, Sustainable AI
Learn more at sambanova.ai/solutions/argyll