Summary
- Ceres estimates data centres could account for around half of a 22GW annual SOFC market by 2030.
- Centrica is targeting multi-gigawatt UK and European demand for onsite solid-oxide generation.
- Ceres argues fuel cells can address grid-delay pressures while fitting emerging DC power architectures.
Ceres Power is putting data centre electricity demand at the centre of its solid-oxide fuel-cell growth strategy, arguing that delays in securing grid capacity are opening a market for modular onsite generation across the UK and Europe.
The Horsham-based technology company estimates that the annual global market opportunity for solid oxide fuel-cell systems could reach 22GW by 2030, with approximately half attributable to data centres.
The estimate was included in Ceres’s interim results for the six months to 30 June 2026. Revenue rose 8% to £22.8 million, while its adjusted EBITDA loss narrowed from £11.3 million to £6.8 million.
Ceres does not manufacture complete power plants itself. Its licensing model places the core solid-oxide technology with manufacturing partners, while commercial partners help develop and deliver projects.
That model is increasingly being aimed at data centre and other energy-intensive loads. Ceres has a strategic partnership with Centrica, which it says sees multi-gigawatt demand for onsite power across the UK and Europe as grid-connection delays affect digital, industrial, and commercial projects.
Centrica and manufacturing partner Delta Electronics have also announced an infrastructure partnership intended to provide megawatt-scale grid-independent generation for UK and European data centres and other high-load industries.
Time to power is reshaping backup and primary generation
The opportunity Ceres is targeting sits between the electricity grid and conventional standby generation. New data centres normally require a firm grid connection for primary power and backup systems designed around outages. Where a grid connection cannot arrive on the required commercial timetable, developers are increasingly examining generation that can carry a larger share of normal operating load.
Ceres said grid upgrades can take between five and 15 years in some markets and contrasted that with fuel-cell systems that it believes can be deployed in months once manufacturing and project delivery are established.
Solid oxide fuel cells electrochemically convert fuel into electricity rather than burning it in a conventional engine or turbine. Ceres highlights efficiency, low local emissions, quiet operation, modularity, and a lack of operational water requirement among the technology’s potential advantages.
The fuel source remains important. Ceres’s current platform can operate on natural gas, with the company presenting hydrogen, biogas, and other lower-carbon fuels as future options. Using gas onsite therefore does not remove carbon emissions, even if the conversion process differs from reciprocating engines or turbines.
The technology may also intersect with another change in AI infrastructure: direct-current distribution. Ceres noted that stationary solid oxide systems produce DC natively, while data centre electrical architectures are beginning to examine high-voltage DC distribution including 800V systems.
In principle, fewer AC/DC conversion stages could reduce equipment count, losses, and floor-space requirements. Ceres cited external estimates suggesting the removal of elements of conventional AC infrastructure could lower data centre capital expenditure, although actual savings would depend on the complete electrical design.
Commercialisation remains the key constraint. Ceres’s partner network has to build manufacturing capacity quickly enough to serve a market measured in hundreds of megawatts or gigawatts rather than individual demonstration systems.
Delta is investing around £240 million in a Taiwanese fuel-cell factory with initial shipments expected in 2028, while Doosan Fuel Cell has secured an export contract to supply stacks to Germany-based Reverion. Ceres also said Centrica intends to establish a UK demonstration site, with a medium-term objective of MW-scale gas-to-power deployments within three to five years.
Data centres are therefore becoming an important proving ground for a technology that has historically competed in a much broader distributed-generation market. Grid queues are creating the demand; the next question is whether solid-oxide manufacturing, fuel supply, permitting, lifecycle maintenance, and project economics can scale quickly enough to become a routine part of European data centre power design.
