DISTRIBUTED AI INFRASTRUCTURE
AI infrastructure, closer to where it is used.
CEAC is a community-scale edge data-centre architecture being developed for a more distributed AI future. Each node combines local AI compute, storage, resilient networking, protected power and cooling in a deployable 20ft infrastructure unit designed around approximately 100 households.
The aim is not to replace hyperscale data centres. It is to create a complementary layer for inference and other workloads that do not need to be concentrated in very large facilities.
Hyperscale where concentration is necessary. Distributed infrastructure where it is not.
The infrastructure problem is changing
UK AI demand is rising at the same time that power availability, grid connections and large data-centre delivery are becoming strategic constraints.
~1.8 GW
Current UK data-centre IT load
Across approximately 330 data centres in the DSIT baseline.
6.3 GW
Medium 2030 demand scenario
Against 5.3 GW of supply in the matched medium DSIT scenario.
57–71%
Potential inference share by 2035
DSIT estimates inference could become the dominant share of UK compute demand.
125 GW
Contracted demand offers
Ofgem reported an increase from 41 GW in November 2024 to 125 GW in June 2025.
Together, these trends point towards a UK compute landscape that will need both large centralised facilities and more distributed forms of infrastructure. Sources
A second infrastructure pathway
A large AI campus concentrates compute, power and network capacity in one location. CEAC explores the opposite end of the topology: smaller protected compute nodes distributed across communities and connected as a wider fabric.
LOCAL COMPUTE
Run suitable inference, retrieval, speech, vision and document workloads closer to the users and data that need them.
LOCAL INFRASTRUCTURE
Combine compute with storage, protected power, cooling, connectivity, monitoring and physical security as one serviceable asset.
FEDERATED CAPACITY
Future CEAC nodes are intended to cooperate so workloads can be placed according to locality, available compute, network conditions, energy and resilience.
PROTOTYPE
CEAC is being developed as complementary distributed AI infrastructure — principally for distributed inference, local data processing, local storage, resilient connectivity and privacy/locality-sensitive workloads.
CURRENT BASELINE
Approximately 100 households
One Generation-1 CEAC node
Generation-1 is currently modelled around approximately 100 households, providing a practical scale for community infrastructure, local compute and shared connectivity.
GENERATION-1 MODEL
Explore the platform
CEAC
What one node contains and which workloads fit.
How it works
Local execution, federation and remote escalation.
Infrastructure
Compute, power, cooling and energy profiles.
Network
100-home demand, backhaul and Smart Atom.
Pricing
Current Generation-1 planning economics.
Engineering model
Assumptions, roadmap and technical binders.
Design a planning scenario
Explore how household count, compute, connectivity, energy and lifecycle choices affect a CEAC deployment.
Design your community node