Sixty-One Gigawatts of Factory, Sixteen Gigawatts of Project: The Electrolyser Market Built for Demand That Hasn't Arrived

At the end of 2024 the world had two gigawatts of water electrolysis installed. Between January and July 2025 it added more than a gigawatt on top of that — by a wide margin the strongest half-year the technology has recorded. Now hold those two numbers beside a third. Global electrolyser manufacturing capacity was already close to 38 gigawatts a year in 2024 and is announced to reach 186 gigawatts a year by 2030 — a compound annual rate of 30.3% and 4.9 times the 2024 base, both Pheonix Research calculations from the IEA’s figures in the Global Hydrogen Review 2025.

An industry that can build 38 gigawatts a year and has installed three is not waiting for a technology breakthrough. It is waiting for customers. And the shape of that wait — not the efficiency of a stack, not the cost of electricity — is what now determines who survives in electrolysers.

Two estimates, one direction

The World Bank, in work published in February 2026, put global electrolyser manufacturing capacity at 61 gigawatts a year as of July 2025, with a further 16 gigawatts under construction and alkaline technology accounting for 43 gigawatts of the total. It put installed capacity dedicated to hydrogen production at roughly 2.15 gigawatts by mid-2025.

The two estimates differ because they measure different things — commissioned nameplate capacity against an annual build rate, different survey dates, and different treatment of lines that are announced but not tooled. What matters is that they point the same way, and that the ratio they produce is not close. Take the 61-gigawatt figure and 16 gigawatts of plant under construction, and the industry’s project pipeline is roughly a quarter of the capacity available to serve it. Take the IEA’s 186 gigawatts of announced 2030 capacity against the three gigawatts actually installed today, and capacity represents 1.6% of what the world has said it will be able to build — a Pheonix Research calculation on the IEA’s own numbers.

Either way, the conclusion is the same. Electrolysers are not supply-constrained. They are demand-constrained, and they are demand-constrained at precisely the moment their manufacturing base has become large enough to compete hard on price.

One country holds two-thirds of both columns

China accounts for 65% of global installed electrolysis capacity and 65% of the capacity that has reached a final investment decision, according to the IEA. It holds close to 60% of global electrolyser manufacturing capacity. Its existing manufacturing base, at 20 gigawatts a year, is roughly ten times the level of Chinese demand in 2024, which was around two gigawatts — a Pheonix Research calculation on the IEA’s two figures.

That is the single most important structural fact in the sector, and it is routinely read as a supply-chain risk. It is better read as a price fact. A domestic manufacturing base running at a tenth of its capability, inside the lowest-cost industrial system in the world, does not idle politely. It exports. And the export price it sets becomes the ceiling under every Western stack maker’s margin, regardless of whose membrane performs better in a datasheet.

There is a counter-movement coming, and it is large. The IEA expects China’s share of manufacturing capacity to fall to roughly one third of the global total by 2030, as worldwide capacity expands toward 186 gigawatts a year. The European Union alone reports 8.9 gigawatts a year of electrolyser manufacturing capacity at the end of 2025, with plans for 41.7 gigawatts a year by 2030, according to SETIS and the European Commission. The world is therefore about to build an enormous amount of electrolyser factory outside China — to compete with the cheapest electrolyser factory inside it. That is a recipe for consolidation, not for a supply squeeze.

Why the factory got built before the demand

The sequencing was not an accident of private capital. Between 2020 and 2025, public policy rewarded the construction of manufacturing capacity and the announcement of projects, and it rewarded them on timelines that ran ahead of the offtake contracts that would have made the projects real. Capacity targets were easier to legislate than hydrogen purchase obligations. Subsidy schemes paid for capex. Almost nothing paid for the molecule.

The consequence is a sector whose industrial base was financed by policy ambition and whose revenue depends on final investment decisions — a decision type that is discretionary, slow, and made by a small number of balance sheets. Electrolyser demand is not a demand curve. It is a sequence of gates.

The squeeze on the pure plays

By 2026 that mismatch had started to produce casualties, and the pattern of them is instructive.

Green Hydrogen Systems filed for bankruptcy in June 2025; its core technology assets were subsequently acquired by thyssenkrupp nucera. In early 2026 Cummins signalled that it intended to stop new commercial electrolyser activity at its Accelera unit. The IEA’s own assessment of the period is blunt: manufacturers outside China are experiencing sharp reductions in revenue and increased financial losses, and for some this has led to bankruptcies or acquisitions, in what may signal a coming wave of consolidation.

Note who is buying. Not a startup with a better stack, and not a private equity fund with a thesis about the hydrogen decade. An industrial conglomerate — thyssenkrupp nucera — taking technology from a failed balance sheet because it can carry development risk across a portfolio. The same logic runs the other way up the chain. Linde, at a market capitalisation of roughly USD 226 billion, generates the large majority of its revenue outside hydrogen. Air Products, one of the largest merchant hydrogen suppliers in the world, added a USD 140 million liquid hydrogen infrastructure contract in 2026. These are businesses for which an electrolyser order book is a division, not an existence.

That is the real competitive divide in the sector, and it has nothing to do with membrane chemistry. It separates companies that can absorb three bad years of final investment decisions from companies that cannot. The pure plays — Nel, ITM Power, Plug Power, and the newly consolidated thyssenkrupp nucera among them — are still largely loss-making in their electrolyser activity and dependent on policy support for project economics. Capital is concentrating in diversified industrial gas majors for exactly that reason.

The orders exist. There are simply very few of them.

Against that backdrop, the 2026 order flow looks less like a market recovering and more like a market concentrating.

thyssenkrupp nucera won a front-end engineering design study for a 260 MW green hydrogen and ammonia project in India, and a 300 MW engineering, procurement and supply contract for Moeve’s alkaline electrolysis project in Spain. ITM Power received a notice to proceed in February 2026 to supply PEM electrolysers to a 20 MW green hydrogen project, and reported in August 2026 that its GET H2 Nukleus project had produced and delivered green hydrogen. Next Hydrogen secured USD 3.75 million of contracts to demonstrate electrolyser technology for nuclear applications.

Read the sizes, not the names. When the installed base of the entire world is around three gigawatts, a single 300 MW contract is a tenth of the global fleet. Four or five awards of that scale in a year can move the industry’s annual installation figure by a third. That is a market in which a handful of procurement decisions, taken by a handful of developers, determine whether a manufacturer’s plant runs at 15% utilisation or 40%. It is also why the sector’s most-watched indicator has stopped being the levelised cost of hydrogen and become the rate of final investment decisions.

The materials race is real, and it is not the fix

There is genuine technical work underway, and two constraints are driving it. The first is iridium and platinum. PEM stacks depend on precious-metal catalysts, and the supply chain constraints around those metals add both cost and timing risk to any large PEM order — the World Bank identifies this alongside insufficient renewable power infrastructure as one of the two recurring constraints that slowed near-term deployment.

The responses are rational. Anion exchange membrane architectures aim to avoid precious-metal catalysts entirely and offer a low-cost route around the constraint. Solid oxide electrolysis is gaining traction wherever an operator has waste heat to improve overall system efficiency, and European manufacturers have concentrated innovation in high-pressure PEM and SOEC designs under the Green Deal framework. Conglomerates are forming joint ventures with technology startups to lock supply chains and commercialise high-capacity stacks faster.

All of it matters for the 2030s. None of it fixes the 2026 problem. A cheaper stack lowers the cost of a project that has not been financed; it does not cause the project to be financed. And in the near term, a cost-down race among manufacturers with empty order books compresses supplier margins rather than expanding demand.

What would actually close the gap

Three things would, and only one of them is in the industry’s control.

Offtake, not announcements. A project without a buyer for its molecules is a press release with an engineering appendix. The pipeline is rich in projects and thin in financeable demand, and the gap between the two is where the entire equipment market is stuck.

Policy rules that are stable long enough to underwrite. The United States Treasury’s implementation guidance on the 45V production tax credit — specifically the additionality and time-matching requirements — remains the single most consequential unresolved variable in the largest subsidised market. A rule change does not shift a demand curve gently; it re-prices an entire pipeline in a quarter.

Industrial gas majors and refiners choosing to convert. The most reliable demand in the sector is captive: existing ammonia, methanol and refining capacity that must decarbonise and can sign a fifteen-year offtake with itself. That demand converts at the pace of board approvals, not at the pace of electrolyser cost curves — which is why it is slow, and also why it is bankable.

What to watch

• Utilisation, not capacity. Any new manufacturing announcement above 186 gigawatts a year is a data point about ambition, not about the market. Watch the ratio of installed gigawatts to manufacturing gigawatts instead, and watch whether it climbs off 1.6%.
• Western order books against Chinese prices. Whether a European or American stack can win a competitive tender in a third market without a domestic-content rule attached is the test that matters. If it cannot, the consolidation has barely started.
• Iridium intensity. A measurable fall in iridium loading per megawatt in commercial PEM product — not in a laboratory paper — is the signal that the materials constraint is loosening.
• Final investment decisions in India and the Gulf. The 260 MW Indian ammonia project is a FEED study, not a purchase order. Watch whether it converts, because a conversion in a low-cost-power geography is worth more to the global installed base than another European subsidy round.
• Who buys the next distressed manufacturer. If the answer remains industrial gas majors and diversified conglomerates, the sector has accepted that electrolysers are a division, not a company.

The bottom line

The electrolyser industry has built the factory the 2030s will need and is running it for a market that has not arrived. Sixty-one gigawatts a year of manufacturing capacity against three gigawatts installed; sixteen gigawatts of plant under construction against a project pipeline that has to clear a final investment decision before a single stack is ordered; 65% of both deployment and committed capacity in one country whose own plant runs at a tenth of its capability.

Those are not the markings of a supply chain being stressed by growth. They are the markings of a manufacturing sector that has been funded ahead of its customers, and is now being consolidated by the only kind of owner that can wait: the one whose balance sheet is not made of electrolysers.

For buyers, that is good news. Stacks will be cheaper in 2028 than in 2026, because too many competent manufacturers will be chasing too few orders. For sellers, the strategic question has already changed. It is no longer whether the hydrogen economy arrives. It is whether the company selling the machine is still independent when it does.

References

International Energy Agency — Global Hydrogen Review 2025. Installed water electrolysis capacity, China’s share of installed and FID capacity, manufacturing capacity, and the assessment of manufacturer finances outside China.

World Bank, February 2026 — global electrolyser manufacturing capacity as of July 2025, capacity under construction, technology split, installed dedicated hydrogen production capacity, and iridium and platinum supply constraints.

SETIS / European Commission, January 2026 — European electrolyser manufacturing capacity and 2030 plans.

thyssenkrupp nucera, March 2026 — 260 MW FEED study for a green hydrogen and ammonia project in India; 300 MW engineering, procurement and supply contract for Moeve’s alkaline electrolysis project in Spain.

ITM Power / Hydrogen Europe, February 2026 — notice to proceed for PEM electrolysers on a 20 MW green hydrogen project; ITM Power, August 2026 — green hydrogen produced and delivered at GET H2 Nukleus.

Next Hydrogen / FuelCellsWorks, March 2026 — USD 3.75 million of contracts for electrolyser demonstration in nuclear applications.

Green Hydrogen Systems — bankruptcy filing, June 2025; technology assets acquired by thyssenkrupp nucera. Cummins Inc. — statement on new commercial electrolyser activity at Accelera, early 2026.

US Department of the Treasury — implementation guidance on the section 45V clean hydrogen production tax credit, including additionality and time-matching requirements.

Pheonix Research — Energy & Power industry coverage.

Frequently Asked Questions

How much electrolyser manufacturing capacity exists globally?

Two independent estimates put it in a narrow band. The IEA reports close to 38 gigawatts a year in 2024, rising to an announced 186 gigawatts a year by 2030. The World Bank reports 61 gigawatts a year as of July 2025, with a further 16 gigawatts under construction and alkaline technology accounting for 43 gigawatts. The measures differ — annual build rate against commissioned nameplate, and different survey dates — but both describe a manufacturing base many times larger than current installation.

How much electrolyser capacity has actually been installed?

Around three gigawatts. Global installed water electrolysis reached 2 gigawatts in 2024, and more than 1 gigawatt was added between January and July 2025. The World Bank puts installed capacity dedicated to hydrogen production at roughly 2.15 gigawatts by mid-2025. Against 186 gigawatts of announced 2030 manufacturing capacity, that is 1.6% — a Pheonix Research calculation.

Why is China's share of the electrolyser market so important?

China accounts for 65% of global installed electrolysis capacity and 65% of the capacity that has reached a final investment decision, and holds close to 60% of global manufacturing capacity. Its own plant, at 20 gigawatts a year, is roughly ten times Chinese demand of around two gigawatts in 2024. A low-cost manufacturing base running at a tenth of its capability exports, and the price it sets becomes the ceiling under every competitor's margin.

Which electrolyser manufacturers are most exposed to the demand gap?

The pure plays. Nel, ITM Power, Plug Power and the newly consolidated thyssenkrupp nucera remain largely loss-making in electrolyser activity and dependent on policy support for project economics. Green Hydrogen Systems filed for bankruptcy in June 2025 and Cummins signalled in early 2026 that it would stop new commercial electrolyser activity at Accelera. Capital is concentrating in diversified industrial gas majors such as Linde and Air Products, which can absorb development risk because hydrogen is a division rather than the whole business.