The reactor is becoming modular; the plant is not
Electrolysers scale by repeating cells, which resembles manufacturing more than building a giant furnace. But a stack still needs clean feed, membranes, electrodes, power electronics, cooling, purification, and a product market.
Feed and electricity
Water, nitrogen, CO₂, salts, or intermediates must arrive at controlled purity, pressure, carbon intensity, and price.
- Measure
- $/MWh · purity · pressure
- Failure boundary
- Clean power availability and feed contaminants both gate the cell.
Where the frontier moves
Cheaper firm clean power and feed pretreatment that tolerates wider input quality.
Cell and stack
Catalysts, electrodes, membranes, electrolyte, flow fields, cooling, compression, and rectification convert charge into bonds.
- Measure
- V · A/cm² · stack life
- Failure boundary
- Overpotential and degradation eat into both efficiency and component life.
Where the frontier moves
Longer stack life and higher current density without sacrificing selectivity.
Selectivity and separation
Faradaic yield, concentration, recycle, drying, and purification determine saleable output.
- Measure
- Faradaic efficiency · product wt%
- Failure boundary
- Dilute or mixed streams can consume more energy in separation than the reactor itself.
Where the frontier moves
In-cell concentration and membranes that do separation's job for free.
Storage, logistics, and demand
The molecule must be buffered, transported or used onsite, certified, and sold under contracts that support plant finance.
- Measure
- Delivered $/kg · firm offtake
- Failure boundary
- Infrastructure and premium buyers are scarcer than announced production capacity.
Where the frontier moves
Co-location with existing demand and long-term offtake contracts.
Charge is conserved and conversion losses compound
Faraday's law fixes electrons per mole; thermodynamics fixes a minimum potential. Real plants add activation, resistance, mass transfer, pumping, drying, compression, and separation. Converting the product back to power compounds the penalty.
What does a kilogram of electrolytic hydrogen actually cost?
The model isolates electricity, capital recovery, and O&M for the simplest electrochemical product. Carbon- and nitrogen-based molecules inherit this foundation and add selectivity and separation losses on top.
Plant-gate production cost
- Electricity
- $3.30 · 69%
- Capital recovery
- $0.94 · 20%
- Operations & maintenance
- $0.57 · 12%
- Thermodynamic floor (HHV)
- $2.36
Hydrogen's higher-heating-value energy content is 39.4 kWh/kg, the thermodynamic floor shown assumes 100% conversion, which no real plant reaches. The utilization slider is the one worth moving first: at 40% utilization a cheap tariff can still produce expensive hydrogen because capital and O&M are spread across far fewer kilograms per year.
Calculation and boundaries
$/kg = electricity price × kWh/kg ÷ 1,000, plus capex ÷ (kg produced per rated kW per year × 15-year life), plus 4%-of-capex annual O&M ÷ kg per rated kW per year. Capex range follows IEA's Global Hydrogen Review 2025 assumptions ($620–960/kW 2030 base case; $1,500–2,400/kW for a Chinese electrolyser landed outside China with tariffs). Excludes compression, drying, storage, transport, financing structure beyond simple amortization, and any carbon or policy premium.
The same electrolyser costs different amounts in different places
Equipment learning curves are real, but engineering, procurement, construction, tariffs, and financing dominate the delivered capital number more than the stack itself does.
What a rated kilowatt of electrolysis costs
None of this is a delivered $/kg figure by itself. As the calculator above shows, a cheap electrolyser running at low utilization can still produce expensive hydrogen, because fixed capital and O&M are divided across fewer kilograms of annual output.
Cheaper reactors move scarcity into utilization and markets
Once a catalyst works, bankability depends on thousands of hours, replaceable stacks, impurity tolerance, concentrated output, shared infrastructure, credible carbon accounting, and buyers willing to contract above incumbent fossil cost.
Start with existing demand
Replace fossil hydrogen and reagents in ammonia, refining, and established chemical chains before inventing new distribution.
Raise useful current
Improve voltage, current density, selectivity, concentration, and component life together, not one at a time.
Integrate the plant
Use heat, recycle, storage, and flexible operation without starving continuous downstream units.
Contract the premium
Standards, procurement, quotas, and carbon policy convert environmental value into revenue a plant can finance against.
Who is building what
Proton exchange membrane (PEM), pressurized alkaline, high-temperature solid oxide (SOEC), and anion exchange membrane (AEM) electrolyzers turn electrons into chemical bonds. Search the record, or filter by stack architecture.
Nel HydrogenA-Series & M-Series (Alkaline & PEM)High-pressure alkaline and advanced PEM electrolyzer stacks manufactured on fully automated production lines in Herøya, Norway
- Reported evidence
- Over 3,500 systems installed worldwide across industrial gas, fueling, and power-to-ammonia projects.
- Announced next step
- Scaling gigawatt manufacturing capacity to push delivered stack capital expenditure below $300/kW.
- Unresolved risk
- Dynamic response limitations of pressurized alkaline stacks during rapid wind/solar fluctuations and balance-of-plant footprint.
ITM PowerTRIDENT & NEPTUNEMulti-megawatt modular PEM electrolyzers engineered for fast ramp rates, high current density, and integration with dynamic renewable power
- Reported evidence
- Supplied the 10 MW PEM system at Shell's Rhineland Refinery; standardized 20 MW plug-and-play modules under commercial manufacture.
- Announced next step
- Standardized 100 MW+ modular balance-of-plant systems optimized for low footprint and rapid onsite installation.
- Unresolved risk
- Iridium catalyst supply constraints on the anode and platinum loading costs under global terawatt-scale deployment.
Thyssenkrupp Nucerascalum 20 MW ModuleStandardized 20 MW alkaline water electrolysis modules based on chlor-alkali cell heritage, optimized for multi-hundred-megawatt gigaprojects
- Reported evidence
- Selected for landmark multi-gigawatt projects, including the NEOM Green Hydrogen Project (2 GW) in Saudi Arabia and Stegra in Sweden.
- Announced next step
- Delivering multi-gigawatt annual electrolyzer capacity with proven 80,000+ hour operational stack lifetimes.
- Unresolved risk
- Lye (KOH) pump and circuit maintenance, caustic corrosion over multi-decade operations, and larger footprint than PEM.
Bloom EnergyBloom Electrolyzer (SOEC)High-temperature solid oxide electrolyzer cells operating at ~800°C using steam and industrial waste heat to achieve >85% electrical efficiency
- Reported evidence
- Demonstrated commercial operation at NASA Ames and industrial chemical sites; produces ~20–30% more hydrogen per kWh than PEM or alkaline when steam is supplied.
- Announced next step
- Direct thermal coupling to nuclear power plants, steel mills, and ammonia synthesis loops to exploit high-grade heat.
- Unresolved risk
- Thermal cycling degradation of ceramic cells during sudden power shutdowns and brittle ceramic seal cracking.
TopsoeSOEC Manufacturing PlantHigh-temperature solid oxide electrolysis cells designed for industrial-scale Power-to-X production of green ammonia and e-fuels
- Reported evidence
- Constructing commercial 500 MW SOEC manufacturing facility in Herning, Denmark; operational pilot stacks demonstrate high syngas co-electrolysis efficiency.
- Announced next step
- Commercial commissioning of the Herning gigafactory, supplying industrial decarbonization projects across Europe.
- Unresolved risk
- High stack capital cost relative to mature alkaline systems and specialized high-temperature alloy piping and insulation.
EnapterAEM Nexus & MulticoreAnion Exchange Membrane (AEM) technology combining the low cost of alkaline materials (no noble platinum/iridium catalysts) with the high current density of PEM
- Reported evidence
- Commercial modular units deployed in over 50 countries; commissioned first multi-megawatt AEM Multicore system in Germany.
- Announced next step
- Displacing both PEM and alkaline systems through ultra-low stack materials costs and standardized modular manufacturing.
- Unresolved risk
- Membrane alkaline durability and mechanical stability under high differential pressures over multi-year duty cycles.
Electric Hydrogen (EH2)100 MW PEM PlantHigh-current-density, high-pressure PEM electrolysis systems engineered from first principles as complete 100 MW industrial plant packages
- Reported evidence
- Commissioned industrial manufacturing facility in Devens, Massachusetts; secured multi-hundred-megawatt commercial customer reservations.
- Announced next step
- Delivering delivered hydrogen at $1/kg by slashing balance-of-plant civil, structural, and electrical installation costs.
- Unresolved risk
- Managing extreme heat flux at ultra-high current densities and proving long-term membrane thinning resistance.
International Energy Agency (IEA)Global Hydrogen ReviewGlobal tracking and technical benchmarking of clean hydrogen projects, stack degradation rates, levelized costs, and policy mandates
- Reported evidence
- Annual published reports documenting project realization rates, tracking that less than 10% of announced green hydrogen capacity reaches final investment decision (FID).
- Announced next step
- Establishing harmonized international standards for life-cycle carbon accounting and certification of clean hydrogen.
- Unresolved risk
- Widening gap between announced project pipelines and actual capital commitments due to power purchase agreement pricing.
Electrolyzer stack efficiencies (~60–80% LHV) exclude balance of plant (rectifiers, water deionization, hydrogen compressors, dryers) which consume another 5–15% of delivered electricity.
The optimistic view, with conditions
Electrons become ordinary chemical feedstock
Electrochemistry wins first where the incumbent is carbon-intensive, clean electricity is structurally cheap, existing demand is concentrated, and cells produce a durable, concentrated stream that fits a repeatable plant.
Displace fossil hydrogen where pipes already exist
Ammonia, refining, and methanol are the logical first markets because handling and continuous demand already exist.
Let utilization, not the stack, decide cost
As equipment cost keeps falling toward IEA's 2030 base case, delivered cost increasingly tracks capacity factor and power price.
Extend the same accounting to harder molecules
CO₂ conversion, ammonia, and methanol inherit hydrogen's cost structure and add selectivity and separation on top.
What a cheap electrochemical molecule actually needs
- High utilizationEnough operating hours per year to spread capital and O&M across a large annual output.
- Durable, selective cellsLong stack life at high Faradaic efficiency and current density together, not traded against each other.
- Cheap, firm powerElectricity priced and delivered on a schedule that matches the plant's duty cycle, not just an annual average.
- Concentrated separationMembranes and process design that avoid paying twice: once in the reactor, once in purification.
- Contracted offtakeBuyers willing to pay above incumbent fossil cost, underwritten by policy or standards.
Which parts of chemistry can electricity replace?
Electrification is already routine in chlor-alkali production and aluminium smelting; their existence establishes technical feasibility for particular reactions, not for all molecules. IEA estimates more than half of chemical-sector energy inputs are feedstocks, the atoms kept in products. An electric heater can remove furnace fuel, but it cannot supply the carbon or hydrogen in methanol or polymers. Ammonia can use electrolytic hydrogen, and methanol can use electrolytic hydrogen plus a non-fossil carbon source; both still need separation, synthesis and a buyer. BASF, SABIC and Linde opened an electrically heated steam-cracker demonstration in 2024, which addresses heat rather than replacing the hydrocarbon feedstock.
The energy convention matters. DOE’s electrolysis handbook distinguishes hydrogen HHV at 39.42 kWh/kg from LHV at 33.31 kWh/kg. The reversible cell voltage near standard conditions is about 1.23 V; 1.48 V is the thermoneutral voltage that supplies both electrical and reaction-heat requirements electrically. Thus 39.4 kWh/kg is not a universal minimum electrical draw if heat enters from outside the stack. Published efficiencies must state HHV or LHV, stack or plant, and whether compression is included.
IEA counted 2 GW of installed water electrolysis in 2024; its 2026 production review puts capacity above 4 GW after 2025, while low-emissions hydrogen output remained below 1 Mt against more than 100 Mt of total demand. A separate IEA project audit contrasts nearly 27 Mt/yr of announced 2030 low-emission hydrogen with just over 4 Mt/yr linked to projects with investment decisions. Electrolysis capacity, annual hydrogen output and chemically useful offtake are different curves.
The conditional economic answer is simple: at 50 kWh/kg plant electricity use, each $10/MWh change in power price changes hydrogen cost by $0.50/kg before capital and delivery. At $30/MWh power alone costs $1.50/kg; at $80/MWh it costs $4.00/kg. Those are arithmetic sensitivity points, not parity prices. Gas price, plant utilization, capital recovery, carbon intensity and transport set whether a given ammonia or methanol project can displace the fossil route. The industrial heat report covers the separate furnace-heat problem.
Project attrition and catalyst supply set two limits
The IEA's 2026 hydrogen review reduced projects with a strong chance of 2030 operation from about 10 Mt/yr to just over 6 Mt/yr after investment delays; 22 Mt/yr of announcements could miss 2030 without decisions by early 2027. In 2025, China installed nearly three quarters of new electrolyser capacity. These are project and capacity measures, not operating chemical output.
IRENA estimates global iridium production around 7–7.5 tonnes a year; its illustrative 400 kg/GW PEM assumption would exhaust that flow at about 18 GW/yr if all metal went to new PEM stacks. Recycling, lower loading and alkaline or solid-oxide systems can change the limit. DOE targets lower total platinum-group loading while requiring durability, so iridium demand per GW is a moving engineering variable rather than a fixed physical constant.
Sources, method, and boundaries
IEA figures describe measured hydrogen production, capex assumptions, and project status. Faraday's law and hydrogen's heating value establish physical frames used throughout. Laboratory cell metrics are treated as component evidence, never as delivered plant cost. The interactive model is a derived, illustrative levelized-cost identity, not a project quote.
- Faradaic efficiency
- The share of electrical charge that produces the target molecule rather than side reactions.
- HHV
- Higher heating value, the total energy released including water vapor condensation, used as hydrogen's thermodynamic reference.
- Balance-of-plant
- Everything around the electrochemical cell itself: pumps, compressors, purification, cooling, and controls.
