Summary
- A 72MW data centre is proposed at Thames Valley Park, replacing underused office buildings on Microsoft’s former Reading campus.
- Consultation material says the project would use fuel cells as an on-site energy solution while additional grid capacity is brought forward.
- The scheme captures the UK’s speed-to-power dilemma: digital infrastructure demand is moving faster than grid reinforcement.
A 72MW data centre proposed for Thames Valley Park in Reading would use on-site fuel cells while additional grid capacity is brought forward, placing gas-backed speed-to-power at the centre of a UK planning debate.
The proposed development would replace underused office buildings on the former Microsoft campus at Thames Valley Park. Project consultation material says the site would support digital infrastructure demand, deliver more than £200m of construction investment, and create up to 245 construction jobs and at least 115 full-time equivalent roles once operational.
The project brochure says the development would use fuel cells as an on-site energy solution while additional grid capacity is brought forward. The fuel cells would use natural gas to generate electricity through a chemical reaction rather than conventional combustion, and the project team says they would be future-fuel ready for biomethane or hydrogen as those fuels become more widely available.
The same material says the data centre would use air-cooled chillers to minimise water consumption, assess heat reuse opportunities for nearby businesses, and deliver at least 10% biodiversity net gain.
Speed-to-power enters planning
The Reading proposal captures the UK data centre market’s current tension. Demand for capacity is rising around AI, cloud, and enterprise workloads, while grid connection timelines can run beyond the commercial patience of developers and customers.
On-site generation is one response. Fuel cells, gas engines, turbines, batteries, and hybrid systems are all being assessed as ways to bridge the gap between immediate data centre demand and slower grid reinforcement. That approach can make projects deliverable, but it shifts the debate from grid availability to fuel source, emissions, air quality, resilience, and long-term decarbonisation.
Fuel cells can have advantages over conventional combustion in some applications. They can be modular, quieter, and lower-emission at the point of use than diesel or gas-engine alternatives. They also provide a more continuous generation option than standby generators. If they run on natural gas, however, the project still depends on fossil fuel supply unless and until lower-carbon fuels are available at scale and suitable cost.
The future-fuel claim therefore needs precise treatment through planning and operation. Hydrogen and biomethane pathways may improve the long-term carbon case, but planning decisions are made on specific projects, not future fuel markets. Operators will need to show how and when fuel switching could happen, what infrastructure would be required, and how emissions are treated in the interim.
The Thames Valley Park site sits in one of the UK’s main digital infrastructure corridors, with strong enterprise demand, proximity to London, and existing technology-sector land uses. It is also a region where power, planning, and local environmental concerns are becoming more visible.
Cooling and heat reuse join the case
The project’s air-cooled chiller approach reduces water consumption, which is important in a region where water stress and public concern over resource use can influence planning. Air cooling brings its own design trade-offs, including fan energy, acoustic treatment, and performance during hotter weather.
Heat reuse is being assessed. That could strengthen the local case if nearby users can take consistent heat through a viable network. In the UK, data centre heat reuse remains difficult because many sites are not located next to ready heat demand, and distribution infrastructure often has to be funded separately from the data centre itself.
The former office campus setting gives the project a redevelopment argument. Reusing underutilised commercial land for digital infrastructure can be more attractive than greenfield development, particularly where access, utility corridors, and business-park context already exist. The energy profile of a data centre is still fundamentally different from offices.
A 72MW load is a material local infrastructure decision. It will require clear answers on grid reinforcement, gas supply, fuel-cell operation, backup power, cooling noise, heat rejection, transport during construction, and long-term emissions.
The proposal is likely to be watched beyond Reading. Developers are increasingly looking for ways around connection delays, while planning authorities are being asked to judge energy systems that sit somewhere between data centre power plant, transition bridge, and permanent fossil-backed load.
If approved, Thames Valley Park could become a template for gas-backed speed-to-power in UK digital infrastructure. If contested, it will show how quickly the public debate moves from whether data centres are needed to what exactly will power them while the grid catches up.
