Once you know how water electrolysis works, the next question is who builds it. This is a map of the manufacturers by technology, and a reading of what the list itself tells you about where the industry is.
There is no shortage of explanations of how water electrolysis works. There is much less written about who builds the machines, which is the question that comes next for anyone actually specifying one.
This is a working map rather than a ranking. Companies are grouped by the technology they build, because that is what determines whether a supplier belongs on your shortlist at all. Everything attributed to a company here is something that company or a public agency has stated.
First, the shape of the market
The number worth holding onto is not total installed capacity. It is the distance between what was announced and what was built.
Against roughly 320 GW of electrolyser capacity announced for 2030, about 28 GW has reached a final investment decision, and installed capacity stands above 4 GW. Roughly 9% of the announced pipeline has been committed to, and around 1% of it exists.
The manufacturing side shows the same gap from the other direction. Manufacturing capacity is approximately 58 GW per year while actual output is under 5 GW, putting utilisation below 9%. Factories were built for the announced pipeline rather than for the orders that followed.
Read as a sequence, the last six years are legible: national hydrogen strategies in 2020, mega project announcements from 2021 to 2024, and from 2025 a refocus onto projects that can actually be executed. Figures here are from the IEA Global Hydrogen Review 2026, and the under-5 and above-4 notations are the IEA's own.
That gap is the context for everything below. A supplier list assembled in 2026 is a list of companies operating a fraction of the capacity they built, waiting on projects that were announced but not committed.
Alkaline (AWE): the default for large, steady plants
Alkaline electrolysis has been in industrial use since the nineteenth century, and that record is why it still wins most large projects. It runs on a liquid alkaline electrolyte and does not require platinum group metals, so the cell materials are inexpensive at scale.
thyssenkrupp nucera (Germany) is the reference supplier at gigawatt scale, built on decades of chlor-alkali cell manufacturing. McPhy (France) has focused on European industrial and mobility projects. The largest volumes, however, come from the Chinese manufacturers — LONGi, PERIC and Sungrow among them — which is also where most of the world’s electrolyser manufacturing capacity sits.
What you accept with alkaline is a floor on turndown and a slower response to changing power. Those limits are the reason the other three technologies exist.
PEM: chosen for response and pressure
Proton exchange membrane electrolysers use a solid polymer membrane and run at high current density with fast response to changing input, which is why they are specified for variable renewable power and for applications that need pressure and purity at the stack outlet.
Plug Power (United States, founded 1997) came to electrolysis from fuel cells and now spans production, liquefaction and refuelling; its supply of a 100 MW PEM plant to Galp's Sines refinery in Portugal is the largest PEM electrolysis project in Europe. ITM Power (United Kingdom) and Siemens Energy (Germany) are the other established European names, and Nel (Norway) builds both PEM and alkaline lines.
The open questions on PEM are not about performance. They are iridium, whose annual primary supply is on the order of 7 to 8 tonnes with more than 80% coming from a single country, and the 2023 ECHA proposal to restrict PFAS, on which a European Commission decision is expected around 2027. Both sit outside any project's control.
On a fifteen year horizon, an unresolved regulatory question about a core material is a line item, not a footnote.
AEM: alkaline chemistry in a membrane cell
Anion exchange membrane electrolysis puts an alkaline environment inside a membrane cell. The intent is to keep the compact, pressurised, fast-responding cell architecture of PEM while removing the requirement for platinum group metals on the oxygen side.
Enapter (Italy and Germany, founded 2017) is the most widely deployed AEM manufacturer, having supplied more than 5,000 AEM electrolysers to over 360 customers across 55 countries, built around small modules combined into larger systems. Evoloh (United States, founded 2020) takes the opposite approach, building large single stacks, and began production in 2025 at a Massachusetts plant with a stated annual capacity of 3.75 GW.
We build AEM as well, from the electrode and the stack upward, and supply to research groups and industrial customers in 12 countries. Our own longest continuous durability run is 2,281 hours, and our systems operate at up to 10 barg.
AEM is the youngest of the four in commercial terms. The honest open questions are membrane durability over long runs and the depth of the supply base, and any supplier who tells you otherwise is selling rather than answering.
SOEC: highest efficiency, if you have the heat
Solid oxide electrolysis runs above 700 °C and splits steam rather than liquid water. Because part of the energy of dissociation is supplied as heat rather than electricity, electrical efficiency is the highest of the four — provided a source of high grade waste heat is already on site.
Topsoe (Denmark), Sunfire (Germany) and Bloom Energy (United States) are the principal suppliers. The constraint is not the cell but the application: without co-located industrial heat, the case weakens quickly, and thermal cycling remains the durability question.
What the list says
Read together, three things stand out.
- Technology choice is decided by the site, not by a ranking. Load profile, outlet pressure and scale narrow the field before a single specification sheet is compared.
- Capacity is not the constraint. With manufacturing running below 9% utilisation, no project is waiting on a factory; it is waiting on a final investment decision.
- The list is shorter than it was. In February 2026 Cummins announced through its Accelera business that it would stop taking new electrolyser orders, after recording an impairment of about USD 458 million for 2025, citing slower than expected market growth and policy uncertainty. Existing contracts continue.
That last point is worth sitting with. A century old industrial manufacturer with an established PEM line and multi-megawatt deliveries concluded that the near term market did not support new orders. It is a data point about the pace of the market rather than about the technology.
If you are specifying an electrolyser now, the question that survives all of this is not which supplier is largest. It is which supplier will still be answering the phone when the stack needs its first membrane replacement.
Looking at AEM for a specific application?