The Factories Were Built Against the Wrong Number

An industry running at 5% of nameplate is not an industry waiting out a soft quarter. It is an industry whose manufacturing base was sized against a different number than the one that governs orders.

• Cumulative installed electrolyser capacity worldwide is about 3 GW — 2 GW by 2024 plus more than 1 GW added through July 2025. Global factory capacity is roughly 33 GW a year. The fleet built over the industry's entire history is about 9% of twelve months of output.
• On a recent build rate near 1–2 GW a year, implied factory utilisation is about 4–6%. Heavy manufacturing below roughly a third of nameplate does not wait out a cycle — it consolidates, mothballs lines, or exits, because fixed cost per unit at that throughput is not recoverable at any plausible price.
• China holds nearly 60% of manufacturing and 65% of capacity installed or past FID. Chinese factories are the only ones facing a domestic order book, so consolidation is likely to look less like weak firms losing share and more like the manufacturing base relocating next to its demand.

Announced low-emissions hydrogen production for 2030 stands at up to 37 Mtpa, down from 49 Mtpa a year earlier. More than 4 Mtpa is operational, building or past FID — a conversion rate near 11%.

• Set an 11% conversion rate beside an 11:1 ratio of manufacturing capacity to installed base and the overbuild explains itself. The factories were capitalised against the announcement column of the table; the orders come from the FID column.
• The incidence is the uncomfortable part. A cancelled project writes off development cost, which is small. A factory built against the same number writes off plant, which is not. The announcement pipeline was optional for those who announced it and binding for those who tooled up to serve it.
• This also makes the manufacturers leveraged to the conversion rate rather than to the technology. Doubling announcements does almost nothing at 11%; moving 11% to 25% more than doubles orders with no new announcements. The metric that matters is the count of projects reaching financial close — published, lagging, and almost never quoted as the sector's headline number.

Chinese equipment installed in China runs USD 600–1,200/kW. The same equipment installed elsewhere runs USD 1,500–2,400/kW after transport and tariffs, against USD 2,000–2,600/kW for non-Chinese kit.

• At midpoints the domestic advantage of USD 1,400/kW narrows to about USD 350/kW at delivery — roughly 85% of the arbitrage consumed in transit, with the ranges overlapping substantially.
• Neither party gets what the policy intended. The developer outside China pays about USD 1,050/kW more than a Chinese developer for identical equipment, while the Western manufacturer is still undercut by roughly 15% at the point of sale.
• This gap is the only part of the electrolyser cost stack that can move by decision rather than by learning, which makes trade policy the largest short-term lever on installed hydrogen capex outside China — and, at 5% utilisation, means demand-side subsidy risks paying for idle capacity twice.