Concerns about power supply and the lack of sites for new server farms are casting a shadow on the sustainability of the UK datacentre market. Demand for computing power, especially in London, is outpacing what’s available.
In August 2024, real estate consultancy CBRE reported that vacancy rates in key cities—Frankfurt, London, Amsterdam, Paris, and Dublin—hit a record low in the first half of the year. This marked five years of steadily declining vacancy rates in these areas. Notably, London was the only city to see new datacentre capacity added in the second quarter of 2024, with a new 30MW facility opening in West London.
Colocation and hyperscale providers are scrambling for spare capacity to meet the growing needs for cloud and AI services. More facilities like the one in West London are crucial. Since the Labour government took office in July 2024, it has lowered planning permission barriers to encourage datacentre developments in London and the South East. Their manifesto even emphasized the economic benefits of a thriving datacentre industry.
The new government reversed previous decisions blocking large datacentres from being built on protected green belt land, opening up significant tracts of undeveloped land for development.
The government’s view is clear: allowing construction on protected greenspaces is a necessary trade-off for the growth of the datacentre market. Yet, EdgeNebula, a UK startup, suggests an innovative solution. They propose converting unused commercial properties and offices into micro-datacentres. CEO Peter Hannaford envisions linking these sites to create a network of compute capacity to support cloud and AI workloads without needing new buildings.
Hannaford remarks, “We’re not turning entire buildings into datacentres. We’re leveraging existing spaces and power.” According to Synergy Research Group, the demand for generative AI is driving the average size of hyperscale datacentres upward, with projections indicating that their size could double in four years. This escalates the challenge for operators already struggling to find sizable sites.
Hannaford believes EdgeNebula’s approach addresses this challenge by utilizing existing sites, which cuts down the environmental impact compared to new construction. He points to Microsoft’s October 2023 initiative to reduce concrete use in its datacentres as part of its goal to become carbon-negative by 2030.
By using existing buildings, EdgeNebula avoids the heavy carbon footprint tied to new construction. Plus, they’re targeting sites already connected to local power grids, speeding up the conversion of unused offices into functioning datacentres.
“Some operators wait years for regulatory approval on power supplies,” says Hannaford, highlighting that there are ready-to-use power sources sitting idle in many disused commercial spaces. Even with the government pledging to reduce planning barriers, getting approval can still take months. “Repurposing underutilized real estate lets us scale quickly, cut costs, and lessen the impact on nature,” he adds, noting that no special permissions are needed for small setups in existing spaces.
Hannaford launched EdgeNebula in November 2024, drawing on his background in datacentre sector recruitment. His experience taught him the challenges operators face in location sourcing. EdgeNebula plans to handle power, cooling, and connectivity for its sites, entrusting Abu Dhabi-based Maerifa Solutions to supply the hardware needed for cloud and AI workloads.
Since unveiling its concept, EdgeNebula has drawn interest from potential investors and owners of vacant real estate. However, Hannaford is adamant that wholesale conversions aren’t the goal. Analysts predict hyperscale datacentres will need to grow even larger to meet demand for cloud and AI services. Hannaford believes now’s the time to rethink how to deliver these power-hungry solutions, especially with the UK government’s recent action plan for AI opportunities, which proposes designated zones for AI-focused datacentres.
Hannaford criticizes certain large-scale projects, like a planned hyperscale facility in Buckinghamshire, asserting they are unnecessary given the 30 million square feet of vacant offices and industrial properties available across the UK, along with 1,000 identified sites that possess the needed power.
“Our model focuses on sustainable growth for AI, aligning with government priorities while tackling the high startup costs and lengthy timelines of traditional hyperscale approaches,” he asserts. It presents a compelling option for a market eager to expand, leaving everyone watching to see if the industry embraces this innovative proposition.