Hydrogen equipment manufacturing: factories built ahead of orders that may never close

Stack lines and balance of plant are two factories with different economics

Cell and stack production rewards volume: coating, cutting, stacking and compression are repetitive, benefit from automation, and improve on a learning curve. Balance of plant does not. Separators, dryers, compressors, water treatment and power conversion are engineered per project, welded by qualified fabricators, and priced as a construction scope. Firms that treat the two as one operation end up with an automated stack line starved of throughput while their engineers are consumed by piping isometrics. Many operators eventually outsource skid fabrication to keep the capital-intensive half loaded, and defend the stack as the part of the value chain they intend to own.

The input base is narrow in exactly the places you cannot substitute

Membranes come from a handful of producers. Iridium supply is small and concentrated, titanium porous layers are a specialist product, and coated electrode capacity does not expand quickly. On the alkaline route the material set is cheaper but diaphragm supply and nickel electrode processing carry their own constraints. Rectifiers and transformers have their own long queue driven by grid investment elsewhere. Qualifying an alternative catalyst coating is a re-validation exercise involving durability testing, not a switch of purchase order, so procurement leverage in this sector is weaker than the order values would suggest.

Approval attaches to pressure and atmosphere, not to the molecule

What regulators examine is a pressurised system operating in a potentially explosive atmosphere, assessed under national and European pressure equipment and hazardous area regimes with a notified body signing off where the risk category demands it. The practical consequence is timing. Conformity attaches to the assembled installation, so a late engineering change to piping, venting or enclosure design reopens documentation that was closed weeks earlier. Suppliers who standardise a skid envelope and hold it firm ship faster than those who accept developer-driven layout changes during fabrication, even when the change looks trivial on a drawing.

Revenue arrives in milestones, and milestones slip

Contracts are structured like construction contracts: advance payment, drawing approval, factory acceptance, delivery, site acceptance, performance demonstration. Each stage can be delayed by something the equipment maker does not control, including permits, grid connection and the offtaker signing. Meanwhile the supply chain has been committed. Operators protect themselves by phasing material release against milestone receipt rather than against the master schedule, and by pricing cancellation and storage explicitly. The alternative, building to a customer forecast, is how several manufacturers ended a strong year with warehouses full of stacks. Retention held back until performance demonstration is the last exposure, and it can sit unpaid for a long while after the equipment works.

The real exposure is what you promised about degradation

Selling equipment is straightforward; committing to efficiency and stack life over an operating decade is where the balance sheet risk sits. Degradation behaviour under variable renewable input is not the same as behaviour under steady laboratory load, and early fleets are still generating that evidence. A manufacturer that prices aggressively on a learning curve it has not yet achieved, then underwrites performance for a plant it does not operate, has effectively written an option against its own future cost base. Conservative operators cap liability, sell replacement stacks as a planned consumable, and treat service revenue as part of the original economics.

Frequently asked questions

Why does announced electrolyser factory capacity never match what ships?

Announced figures describe nameplate potential of a line if it ran fully loaded on a single product. Actual shipments follow customer projects reaching financial close, which happens later and less often than pipeline announcements imply. Manufacturers also announce capacity partly to qualify for developer shortlists and public funding, since developers screen out suppliers who cannot demonstrate scale. The gap between announcement and output is therefore structural rather than a sign of a specific company failing.

Is it better to make stacks or to integrate complete systems?

They are different bets. Stack production is a manufacturing business where cost falls with volume and intellectual property sits in materials and cell design. System integration is closer to project engineering: lower capital, faster to start, but margin depends on estimating discipline and on managing site risk. A supplier doing both must resist letting project engineering consume the technical staff who should be driving stack cost down, which is the most common failure pattern in the sector.

What causes hydrogen equipment orders to slip so persistently?

Offtake is the usual culprit. A developer needs a buyer committed at a price that supports the project, and industrial buyers have been slow to sign at prices reflecting current production cost. Grid connection queues, permitting for hazardous installations, and subsidy award timetables each add their own delay. Equipment suppliers see none of these directly; they see a customer who stops responding to drawing approval requests, which is why milestone-linked material release matters so much.