The uneven distribution of heat is one of the great ironies of the modern technological age. While households struggle with rising energy caps and the cost of keeping radiators warm, the technology sector is spending billions attempting to vent heat away from data centres. As the demand for Artificial Intelligence (AI) processing power spikes, this thermal byproduct is only increasing.
A pilot scheme in the UK is now attempting to close this loop by moving the data centre out of the industrial park and into the garden shed.
In a trial conducted in Essex, Terrence and Lesley Bridges have replaced their traditional gas boiler with a unit known as a HeatHub. Developed by the startup Thermify, the system is essentially a distributed data centre. It houses approximately 500 mini-computers, reportedly utilising Raspberry Pi architecture, which process data for corporate clients.
The thermodynamics of computing are simple: processing generates heat. In a traditional server farm, this heat is a nuisance that requires energy-intensive cooling systems to manage, often accounting for up to 30% of a facility’s electricity usage according to the International Energy Agency.
In the HeatHub system, this “waste” energy is captured by oil and transferred via a heat exchanger into the home’s hot water system. For the Bridges, the results have been financially dramatic. The couple reports that their monthly energy outlay has dropped from around £375 to between £40 and £60.
The trial is part of the SHIELD project, a collaboration with UK Power Networks and social housing provider Eastlight Community Homes. The initiative aims to find viable paths to net zero for low-income households. Alongside the HeatHub, the property was equipped with solar panels and a battery storage system—technologies that companies like EcoFlow have been popularising for off-grid and hybrid energy independence.
Travis Theune, co-founder and CEO of Thermify, told the BBC that the objective was to solve the problem of providing clean, affordable energy. By decentralising the cloud, the cost of the electricity used to generate heat is effectively subsidised by clients paying for data processing.
While Mr Theune noted that the current iteration of the hardware utilises arrays of small computers roughly the size of a matchbox and is not designed for heavy AI training loads, it is suitable for app hosting and data analysis.
The concept of using compute-heat for warmth is not entirely new, though domestic application has been rare. We have previously seen “digital boilers” used to heat swimming pools in Devon, and companies exploring district heating networks powered by server farms in Cambridgeshire. However, moving the infrastructure directly into a residential setting presents unique challenges regarding maintenance, noise, and internet connectivity stability.
For the geek enthusiast, the idea of living next to a server rack that pays its own way is appealing. It represents a shift from the centralised model of the internet to a distributed physical infrastructure. If the pilot is successful, Eastlight Community Homes intends to roll the technology out to a further 50 homes.
With energy monitoring becoming a staple of the modern smart home, supported by tools like ivie and smart meters, innovations that turn energy consumption into a dual-purpose utility are likely to gain traction.
Microsoft has previously experimented with sinking data centres into the ocean (Project Natick) to utilise natural cooling, and other firms have utilised artificial lakes. Thermify’s approach suggests that rather than sinking the heat into the sea, we should be sinking it into our radiators.
Photo by Mike Kononov on Unsplash.
