AI data centers lift fuel-cell demand

Fuel cells are gaining traction among US hyperscalers building AI-focused data centers that require large volumes of round-the-clock electricity but face constraints connecting to the grid. Linking major new loads to the US grid can take three to six years, pushing developers to consider on-site generation. In contrast, some fuel-cell projects have moved from contract signing to local planning approvals within three months, and power has been delivered to at least one major data-center site in less than two months. This speed, combined with reliability, is making fuel cells an attractive option for AI infrastructure.

Fuel cells generate electricity through a chemical reaction between a fuel such as natural gas or hydrogen and oxygen, converting the fuel’s energy directly into electricity rather than burning it. This approach avoids combustion within the stack and produces lower local emissions than conventional gas-fired generation. For data centers under pressure to add capacity quickly while managing environmental and permitting concerns, this combination of rapid deployment and cleaner local profile is a key driver of demand.

Rystad Energy projects that investments in fuel cells for data centers will rise to about $30 billion by 2030, up from $2.8 billion in 2025. The industry’s current and planned activity is reflected in an estimated order book of roughly 9 GW, including framework agreements with customers such as large technology, utility and infrastructure operators. While such agreements do not all represent firm orders, they highlight the growing role of fuel cells in long-term data-center planning.

Rystad also expects annual fuel-cell manufacturing capacity to increase to 4 GW by 2030 from about 1.8 GW currently. In parallel, it predicts that about 40% of projected US data-center capacity in 2030 could pursue dedicated on-site power rather than rely entirely on the grid. These trends suggest that fuel cells are poised to become an important part of the power mix for digital infrastructure build-outs in the coming years.

Different technology platforms are being positioned for this opportunity. Bloom Energy operates solid-oxide fuel cells that are ready for 100% pure hydrogen, while FuelCell Energy uses molten carbonate technology that can handle hydrogen blends capped at 40%, with both systems described as fuel-flexible. These capabilities align with customer interest in both current natural-gas-based operation and future hydrogen use.

Large cloud and data-center operators are beginning to commit to significant volumes. One US cloud computing company intends to procure up to 2.8 GW of Bloom fuel-cell systems, with 1.2 GW already contracted. For projects of 100 MW or more, company executives highlight that schedule risks tend to stem more from site readiness, permitting and supporting infrastructure such as gas connections, transformers and switchgear than from the fuel-cell technology itself.

FuelCell Energy is increasingly targeting the data-center segment as a central part of its business strategy. The company expects data centers and related activities to represent 75%-85% of its business within two to three years, compared with 25%-30% at present. Its leadership emphasizes that for customers, "time to power" has become as important as the cost of power, underscoring the premium placed on rapid deployment.

Fuel cells can be deployed either as stand-alone on-site generation or as a complement to grid power, and they compete with other on-site technologies. Their growth will depend on whether suppliers can deliver systems at competitive costs and whether developers can build projects and secure fuel supplies. Within that landscape, the AI data-center build-out is emerging as one of the strongest sources of demand for these systems.