Canal-Cooled AI Data Centre
Sustainable, high-density GPU compute cooled by canal water—built to power Media Stream AI, IntuiTV personalisation, and your cloud workloads.
Sustainable by Design
Our canal-side facilities use closed-loop heat exchange with canal water to remove heat from high-density GPU racks, reducing grid strain and fresh-water usage. Waste heat can be repurposed locally for ecology projects and buildings.
- • Canal water heat-exchange with low ecological impact
- • High-density racks for modern GPU clusters
- • Renewable-ready power integration
- • Built for AI training, fine-tuning, and real-time inference
Process & instrumentation diagram (illustrative).
Canal Cooling — 3D Visuals
Restored 3D visuals of the canal-cooled loop and facility integration.
3D overview of canal cooling loop.
Cooling infrastructure integration.
Water-Flow Diagram
Animated schematic showing filtered intake, heat-exchange, and return safeguards.
GPU Clusters & Topology
Sovereign cluster topology with high-bandwidth interconnects and inference edges.
GPU Tiers & Services
Low-latency instances for real-time personalisation and stream metadata.
Mid-range multi-GPU for continual / transfer learning.
High-density multi-node training for vision & recsys models.
Burst rendering for VP / VFX pipelines.
Facility Specifications
| Cooling | Closed-loop canal water heat-exchange + dry coolers |
| Target PUE | < 1.2 (design point) |
| Rack Density | Up to 30–60 kW/rack (high-density aisles) |
| Power | Renewable-ready, multi-feed with UPS + genset |
| Security | 24/7 monitored, multi-factor access, zoned cages |
| Connectivity | Multi-carrier, redundant fiber, private interconnects |
| Compliance | ISO 27001 (planned), SOC 2 (planned) |
Locations & Roadmap
Data Sovereign Cluster (UK). Target: Q1 2026.
Planned EU data centre for sovereign workloads. Target: 2026.
Planned EU data centre for sovereign workloads. Target: 2026.
FAQ
How does the canal cooling work?
A closed-loop heat exchanger transfers heat from data hall water to a secondary circuit interacting with canal water, with filtration and temperature safeguards to protect canal health.
Can we colocate private, sovereign clusters?
Yes. We offer dedicated cages and private subnets with compliance roadmaps for ISO 27001 and SOC 2.
What GPUs are available?
Inference (L4/A10G), fine-tuning (A4000/A5000), training (A100/H100), and media (L40/A6000) tiers depending on workload.
Do you offer managed services?
We provide managed model hosting, inference APIs, burst rendering, and integration with Media Stream AI / IntuiTV.
Need sustainable AI compute?
Get access to canal-cooled GPU clusters or discuss private sovereign builds.
Contact our team