Summary

• Viridor and Rivington Energy plan up to 200MW of AI-ready data-centre capacity across four UK energy-from-waste sites.
• The first proposed 40MW site in Avonmouth could operate by 2027, subject to planning.
• The plan links AI infrastructure growth to local power generation, grid constraints, and planning consent.

Viridor and Rivington Energy are planning up to 200MW of AI-ready data-centre capacity across four UK energy-from-waste sites, in a proposal that connects compute demand with local power generation.

The partnership covers sites in Bristol, Greater London, Oxfordshire, and Fife. Viridor says the facilities would be powered by locally generated electricity from its energy-from-waste operations, supporting UK demand for digital and AI infrastructure. The first proposed 40MW data centre at Avonmouth, Bristol, could be operational by 2027, subject to planning approval.

The plan arrives as the UK tries to expand AI infrastructure while facing familiar constraints: grid connection queues, local planning pressure, energy availability, and concern about whether data centres bring enough local benefit to justify their power use. Energy-from-waste sites offer one possible answer because they already generate power and sit within industrial land-use contexts.

The company’s partnership update frames the model as a way to provide long-term electricity for data centres without relying entirely on the grid. If it works, the approach could reduce pressure on grid connections and make some sites more deliverable than projects waiting for new network capacity.

That does not make the model simple. Energy-from-waste has its own environmental and policy controversies, particularly around waste hierarchy, emissions, and long-term dependence on incineration capacity. Data-centre developers using this power route will need to show not only that energy is available, but that the carbon, air-quality, and local-impact arguments stand up under planning scrutiny.

The proposal also points to a wider change in the data-centre market. AI workloads are pushing developers to look beyond conventional campuses and towards sites where power is the anchor asset. Private-wire arrangements, co-located generation, heat reuse, and grid-adjacent development are all becoming more important as operators search for capacity that can be built quickly enough to match demand.

For the UK, these projects raise a question of infrastructure strategy rather than simple site selection. A compute site connected to local power, industrial land, heat networks, and regional planning could be more credible than a standalone facility competing with housing, factories, and public services for scarce grid capacity. That promise depends on execution rather than site branding.

Planning will be the immediate test. The first Avonmouth project is still subject to approval, and local authorities will need to weigh digital infrastructure demand against environmental impacts, traffic, construction, power use, and community benefit. The broader portfolio will also need customers, grid arrangements, construction funding, and credible operational timelines.

Viridor and Rivington’s plan sits at the point where two policy pressures meet: the need for more UK compute capacity and the need to manage the energy consequences of that build-out. AI infrastructure cannot be separated from the power systems that feed it. Projects that solve for energy, planning, and local industrial use at the same time will have an advantage over those that treat electricity as a procurement detail.