Fuel Cell Market (2026-2036)

The fuel cell market covers systems that generate electricity through an electrochemical reaction rather than combustion, across proton exchange membrane, solid oxide, phosphoric acid, molten carbonate and other cell types, supplied for stationary power, transportation and portable applications and operated on natural gas, hydrogen, methanol and biogas. The growth of this market is driven by demand for electricity at data centre sites that cannot be connected to the grid quickly enough, by national hydrogen programmes in Asia-Pacific and Europe, by the replacement of diesel generation where local air quality rules apply, and by the use of fuel cells in material handling fleets where refuelling time matters more than fuel cost.

Demand in this market divides between stationary power and transport applications, and the two are growing at different rates. Stationary power is growing quickly, because data centre operators need electricity on site within months rather than years and fuel cells running on natural gas can be installed in that time. Transport demand is growing more slowly, because hydrogen refuelling infrastructure remains limited and because the United States has withdrawn the tax credits that supported vehicle purchases.

Most fuel cells sold for stationary power currently run on natural gas or biogas rather than on hydrogen. Customers buy them for the speed and reliability of the installation rather than for a carbon saving. Hydrogen remains the long-term fuel for this market and the systems being installed are built to convert to it, while current sales are made on grid timing and site constraints.

A fuel cell converts fuel into electricity through an electrochemical reaction, which means it has no combustion step and few moving parts. It therefore runs quietly, emits little at the point of use, and holds its efficiency at part load where an engine loses it. These properties decide where fuel cells are bought. They are chosen where electricity is needed on a site that cannot easily be connected to the grid, where local air quality or noise rules restrict engines, or where a vehicle must be refuelled quickly and run for a long shift.

The strongest demand at present comes from data centres, and recent contracts show the scale of it. Bloom Energy announced on 13 April 2026 that Oracle intends to purchase up to 2.8 GW of its solid oxide fuel cell systems under a master services agreement, of which an initial 1.2 GW has been contracted with deployment under way and continuing into 2027. The systems are installed on site as distributed generation rather than drawing on the grid, and the company stated that a previous installation for the same customer was delivered in 55 days against an anticipated 90-day schedule. Bloom Energy also stated that its platform is aligned to emerging standards such as 800 V direct current and is built for denser artificial intelligence workloads, and that the systems provide rapid load-following support that traditional grids were not designed to deliver.

Installation time is the main reason for this demand. A data centre operator waiting for a grid connection measures the delay in years, so a plant that can be installed in under two months allows a site to open that could not otherwise be built. The decision is made on timing and site constraints rather than on the cost of electricity, which is why the orders are large and relatively insensitive to price. The reference to 800 V direct current reflects the same requirement, because the equipment is designed to connect directly into the power system that new artificial intelligence sites are adopting.

Transport and residential demand have moved in the opposite direction following recent changes to United States tax law. The One Big Beautiful Bill Act changed the United States tax treatment of hydrogen and clean vehicles. The section 45V clean hydrogen production credit is terminated for facilities that do not begin construction before 1 January 2028, where the previous schedule allowed construction to begin before 1 January 2033, which brings the cut-off forward by five years. The section 30D clean vehicle credit is terminated for vehicles acquired after 30 September 2025 and the section 45W commercial clean vehicle credit is terminated for qualifying commercial clean vehicles acquired after the same date, in both cases around seven years earlier than the previous schedule. The section 25D residential credit, which covers residential fuel cell systems, ends for systems placed in service after 31 December 2025, while leased residential fuel cell property remains able to claim the section 48E credit until that credit expires.

The effect on hydrogen transport is direct. A buyer of a fuel cell truck or bus in the United States lost the purchase credit for vehicles acquired after 30 September 2025, and a producer of the hydrogen that would fuel it now has until the end of 2027 to begin construction in order to claim the production credit. The two changes work together, because vehicle demand and fuel supply depend on each other, and removing support from both at once leaves each waiting for the other. Hydrogen mobility in the United States is therefore expected to grow from a small base over the forecast period rather than to scale within it.

Residential fuel cells have been affected by the same legislation. The section 25D credit ended for systems placed in service after 31 December 2025, which removes the support that made a home system affordable, although leased residential property retains access to the section 48E credit for the time being. The residential segment of this market is consequently expected to depend on Japan, where domestic fuel cell systems have an established base and a different support structure, rather than on North America.

Grid connection delay accounts for most of the growth in this market. Operators are buying these systems because the alternative is a site that cannot open, rather than because fuel cells are clean, quiet or efficient, although they are all three. Demand is therefore firm while interconnection queues remain long and exposed if those queues shorten, since a utility that reduces its connection timeline removes the reason for the purchase. Suppliers are responding by selling on reliability and load following alongside speed, because those reasons remain valid once grid access improves.

Fuel choice separates this market more clearly than cell technology does. A stationary system running on natural gas is sold on availability and speed, competes with a gas turbine or an engine, and is bought by a site developer. A system running on hydrogen is sold on emissions, competes with a battery or a grid connection, and depends on a fuel supply chain that is still being built. The same manufacturer often supplies both, so the fuel a system runs on describes the business more accurately than the product name does.

Transport demand is concentrated in applications where batteries are least suitable rather than spread across the vehicle market. Material handling is the clearest example, because a forklift refuelled in minutes allows a warehouse to run three shifts, and the fuel is consumed inside a single site where one refuelling point serves the whole fleet. Buses and regional trucks follow the same pattern where routes return to a depot. Passenger cars remain the most difficult application, because a driver refuelling in public needs a station network that does not exist in most markets, and the loss of the United States purchase credit has widened that gap.

Supply is concentrated among a smaller number of companies than the market appears to contain. Solid oxide systems for large stationary sites come from a few manufacturers, proton exchange membrane stacks for vehicles come from a similarly small group, and several of the large industrial names in this market supply components or operate through joint ventures rather than selling complete systems. A buyer placing an order of several hundred megawatts therefore has a short list of suppliers, and delivery capacity rather than product specification usually determines how quickly an order can be filled.

Two conditions qualify the growth rate in this market. The largest orders are concentrated among a small number of customers in a single application, and the data centre demand supporting them rests on a grid connection problem that utilities are under pressure to resolve. If interconnection timelines shorten materially, the market returns to competing on the cost of electricity, where a fuel cell running on natural gas is more expensive than a grid connection in most regions. The forecast in this report assumes queues remain long through the first half of the period and ease gradually in the second.