Levelised Cost of Hydrogen (LCOH)

• Levelised Cost of Hydrogen (LCOH) is a metric that normalizes hydrogen production costs by dividing annualized or discounted lifecycle expenditures by hydrogen output.
• Electricity cost, equipment degradation, and policy credits are key drivers, with studies showing LCOH variations from roughly $0.31/kg to above $6/kg under different system boundaries.
• Temporal dynamics, learning curves, and risk-adjusted financing critically influence LCOH, highlighting the role of operational strategies and public support in cost reduction.

Levelised Cost of Hydrogen (LCOH) is the hydrogen analogue of the levelised cost concepts used in power-system analysis: it expresses the cost of producing hydrogen on a normalized per-unit basis, usually in $$/\mathrm{kg}$,$\euro/\mathrm{kg}$, or$\pounds/\mathrm{kg}$$, by dividing annualized or discounted lifecycle costs by hydrogen output over the relevant horizon. In the literature, however, LCOH is not a single invariant quantity. Its precise meaning changes with the economic boundary, temporal resolution, treatment of electricity procurement, policy credits, financing assumptions, degradation, and whether the reported quantity is an ex post average cost, a weighted system cost, or an optimization-implied hydrogen price (<a href="/papers/1908.10119" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">1908.10119</a>, <a href="/papers/2408.10824" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Atouife et al., 2024</a>, <a href="/papers/2205.11901" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Zeyen et al., 2022</a>).</p> <h2 class='paper-heading' id='formal-definition-and-principal-variants'>1. Formal definition and principal variants</h2> <p>The most classical formulation treats LCOH as an annualized cost divided by annual hydrogen production. In the German heavy-duty hydrogen refueling station study, for example, the metric is written as</p> <p>$$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$</p> <p>where annualized CAPEX and annual OPEX are normalized by annual hydrogen output (<a href="/papers/1908.10119" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">1908.10119</a>).</p> <p>A discounted-cash-flow formulation is also common. In the <a href="https://www.emergentmind.com/topics/per-epoch-noise-mixing-pem" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">PEM</a> electrolyzer degradation study, the metric is defined as</p> <p>$$LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$$</p> <p>so that both costs and hydrogen production are discounted over the plant lifetime (<a href="/papers/2405.06766" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Schofield et al., 2024</a>). This version is structurally close to levelised cost of electricity formulations and is especially useful when stack replacement, long asset lives, and time-varying production are material.</p> <p>Some studies make the unit conversion explicit. The UK wind-powered green hydrogen assessment defines</p> <p>$$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$</p> <p>with$HHV = 39.4$$kWh/kg, so the same result can be reported either per kilogram of hydrogen or per$$\mathrm{MWh}_{\mathrm{H2,HHV}}$$(<a href="/papers/2509.00136" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Andoni et al., 29 Aug 2025</a>).</p> <p>The same label can also denote a policy-adjusted net cost. The U.S. Inflation Reduction Act study evaluates a levelised cost of fuel production for hydrogen and then subtracts <a href="https://www.emergentmind.com/topics/implementation-relaxation-approach-ira" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">IRA</a> subsidies—principally 45V, 45Q, and, for electrolysis, 45Y through the electricity price—to obtain a net LCOH (<a href="/papers/2305.00946" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Cheng et al., 2023</a>). By contrast, the European sector-coupled PyPSA-Eur-Sec study does not present a standalone plant-level LCOH equation; it reports hydrogen prices derived from the optimization, treating them as the LCOH-equivalent output cost of the optimal system configuration (<a href="/papers/2205.11901" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Zeyen et al., 2022</a>).</p> <h2 class='paper-heading' id='cost-structure-and-dominant-sensitivities'>2. Cost structure and dominant sensitivities</h2> <p>A compact decomposition used in the policy-driven electrolysis cost study writes hydrogen cost schematically as</p> <p>$$\mathrm{LCOH} \approx \frac{\text{annualized CAPEX} + \text{fixed OPEX}}{\text{annual H}_2\text{ output}} + \text{electricity cost} + \text{water cost} + \text{compression/storage cost},$$</p> <p>which captures the main terms recurring across most techno-economic studies (<a href="/papers/2408.10824" title="" rel="nofollow" data-turbo="false" class="assistant-link" x-data x-tooltip.raw="">Atouife et al., 2024</a>). Variants add transport, storage turnover fees, stack replacement, liquefaction, battery replacement, or opportunity-cost terms depending on the system boundary.</p> <p>Electricity price is repeatedly identified as the dominant driver once capital costs begin to fall. The policy-driven electrolysis study states explicitly that achieving unsubsidized$$$1$–$\euro/\mathrm{kg}$0 would require unrealistically low electricity costs, and that low-cost hydrogen depends not only on cheaper equipment but also on high utilization, with $\euro/\mathrm{kg}$1–$\euro/\mathrm{kg}$2 utilization supported by dedicated solar or wind alone and $\euro/\mathrm{kg}$3–$\euro/\mathrm{kg}$4 achievable with a mix of resources (Atouife et al., 2024). The UK wind-electrolyser comparison likewise finds that electricity cost is the dominant contributor to LCOH, followed by electrolyser cost, while compressor and interconnection costs are secondary (Andoni et al., 29 Aug 2025).

In system-integrated settings, the electricity term can dominate even more strongly. In the German HDV-HRS study, OPEX accounts for about $\euro/\mathrm{kg}$5–$\euro/\mathrm{kg}$6 of LCOH, and nodal electricity price is the main determinant of the north-south cost gradient across Germany (1908.10119). This is why co-optimization with the power system lowers average hydrogen cost even when it requires larger electrolyzers.

Degradation can alter the cost decomposition materially. The dynamic PEM electrolyzer study reports that including usage-dependent degradation raises LCOH from $\euro/\mathrm{kg}$7 for the 2022 case and decreases stack life to about two years, because higher degraded voltage increases electricity consumption and accelerates stack replacement (Schofield et al., 2024). In off-grid renewable power-to-hydrogen systems, storage and balancing hardware can also become major contributors: in the Inner Mongolia grid-forming BESS study, BESS capital expenditure accounts for $\euro/\mathrm{kg}$8 of total annual cost in the base case, and faster electrolyzer load adjustment reduces both BESS size and LCOH (Zhu et al., 2024).

3. System boundary, market embedding, and interpretive differences

LCOH is highly sensitive to what is included in the modeled system. At the narrowest boundary, it is a plant metric for hydrogen generated by a specific production pathway. At wider boundaries, it becomes a system-integrated economic signal that reflects grid congestion, location, temporal flexibility, and cross-sector interactions.

The German HDV-HRS analysis is explicit that LCOH is defined from a system perspective. Hydrogen is produced locally at each station via electrolysis, but electricity is priced using locational marginal prices from a coupled power-system model; therefore hydrogen cost reflects both station design and where and when electricity is consumed in the German grid (1908.10119). The resulting metric is not simply a generic electrolyzer cost.

The European PyPSA-Eur-Sec study goes further by embedding hydrogen in a continent-scale sector-coupled optimization over 2020–2050 with electricity, heating, transport, and industry. There, the reported hydrogen costs are optimization-derived hydrogen prices, extracted as dual variables after the capacity-expansion problem is solved, so the LCOH-equivalent quantity is effectively the system-optimal marginal cost of hydrogen supply under endogenous deployment and learning (Zeyen et al., 2022).

Other studies use spatially resolved off-grid or regional least-cost models. The global renewable hydrogen production system model optimizes a hybrid hydrogen production system on a worldwide $\euro/\mathrm{kg}$9 km grid with country-specific interest rates and continuous $\pounds/\mathrm{kg}$0 demand (Kigle et al., 2023). The Sub-Saharan Africa cost-potential study uses ETHOS.FINE to construct regional cost-supply curves in which hydrogen demand is increased incrementally in roughly $\pounds/\mathrm{kg}$1 steps, and the resulting marginal cost at each step is reported as LCOH (Ishmam et al., 2024).

A plausible implication is that direct numerical comparison of reported LCOH values is only meaningful after aligning boundary conditions. A plant-level discounted average cost, a system-weighted average cost using nodal prices, and an optimization-derived shadow-price-like hydrogen value are formally related but not identical objects.

4. Temporal dynamics: learning, degradation, replacement, and operational realism

A large part of recent LCOH research concerns temporal dynamics rather than static design. One strand models endogenous or policy-induced learning. The policy-driven emerging technologies study splits electrolyzer CAPEX into stack, balance of plant, and EPC, assigning global learning to stacks and local learning to BoP and EPC. In its base-case regional and technology scenarios, total installed electrolyzer system capital costs are projected to fall by $\pounds/\mathrm{kg}$2–$\pounds/\mathrm{kg}$3 by 2030, driven by gigawatt-scale factories, automation, supply-chain optimization, increasing standardization, larger project sizes, reduced technology risk, and stronger competition (Atouife et al., 2024).

The European sector-coupled learning study formalizes cost decline with an experience curve,

$\pounds/\mathrm{kg}$4

and models local learning for electrolysis and global learning for solar PV and wind (Zeyen et al., 2022). Under endogenous learning, hydrogen costs in the $\pounds/\mathrm{kg}$5 scenario fall to $\pounds/\mathrm{kg}$6 in 2030 and $\pounds/\mathrm{kg}$7 in 2050. The same study reports that omitting dynamic learning-by-doing can overestimate hydrogen cost by up to $\pounds/\mathrm{kg}$8 in 2030 and overestimate total system costs by up to $\pounds/\mathrm{kg}$9.

A second strand centers on degradation and replacement. The dynamic PEM optimization study couples operating current density to stack degradation, which then feeds back into electricity consumption, replacement timing, stack sizing, and storage needs (Schofield et al., 2024). The replacement-strategy study makes the end-of-life threshold itself an optimization-relevant parameter and finds a base-case optimum at a $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$0 degradation threshold with a seven-year replacement period; across degradation-scale assumptions, the cost-optimal replacement time shifts by up to nine years and LCOH minima range from $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$1 to $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$2 (Arnold et al., 22 Aug 2025).

Operational realism also matters. The grid-connected renewable hydrogen planner study shows that full-foresight annual optimization slightly underestimates LCOH relative to day-to-day operation, while underestimating emissions much more strongly; depending on the emissions-weighting setting, day-to-day operation can produce emissions more than $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$3 up to $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$4 higher than the full-foresight benchmark (Farah et al., 2024). This suggests that some low published LCOH values partly reflect optimistic information structures rather than purely technological performance.

5. Policy support, financing conditions, and risk-adjusted discounting

LCOH is strongly shaped by public policy. The policy-driven electrolysis study concludes that enacted policies can materially reduce electrolyzer CAPEX by 2030, but that electrolytic hydrogen at $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$5–$$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$6 would still require policy support; unsubsidized broad competitiveness with fossil hydrogen by 2030 is described as unrealistic because capital recovery and electricity remain too important (Atouife et al., 2024). In that study, the U.S. 45V hydrogen production tax credit is singled out as transformative.

The Inflation Reduction Act analysis makes this mechanism explicit by subtracting levelized subsidy values from plant-level production costs (Cheng et al., 2023). In its early-2030s assumptions, gray hydrogen via SMR has $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$7 and $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$8 when the maximum $$\mathrm{LCOH}=\frac{C^{\mathrm{ann}}_{\mathrm{HRS}}+C^{\mathrm{OPEX}}_{\mathrm{HRS}}}{H^{\mathrm{annual}}},$$9 credit and 45Y-adjusted electricity price are applied. The same paper therefore distinguishes sharply between gross production cost and net policy-adjusted hydrogen cost.

Financing conditions are equally consequential. The global hydrogen production system model introduces country risk premiums into the discount rate, $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$0, with $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$1 and CRPs ranging from $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$2 to $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$3 (Kigle et al., 2023). In the BASE scenario, LCOH ranges from $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$4 to $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$5, with an average of $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$6, compared with an average of $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$7 under a constant WACC.

The natural-hazard discount-rate study extends this logic by adding a natural hazard component to country-specific discounting. It defines a final discount rate $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$8, and also a weighted form with a $LCOH=\frac{PV}{PV_{\mathrm{H_2}}},$9 split between economic and natural-hazard risk in the main scenario (Stargardt et al., 20 Mar 2025). Under alternative discount-rate constructions, the relative difference in hydrogen generation cost ranges from $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$0 in the Philippines to $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$1 in Kyrgyzstan, and comparison with a uniform $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$2 discount rate yields differences from $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$3 in Somalia to $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$4 in Qatar. In this line of work, LCOH is not merely a function of technology and resource quality; it is also a function of how investment risk is priced.

6. Reported values, competitiveness, and recurring interpretive issues

Representative LCOH outcomes span a wide range because the literature spans different currencies, system boundaries, policy environments, and operating concepts.

Context	Economic boundary	Reported LCOH
German HDV-HRS, co-optimized with power system	System-integrated, LMP-based	$$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$5
German HDV-HRS, less system-aware sizing	System-integrated, LMP-based	around $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$6
Europe, $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$7, endogenous learning	PyPSA-Eur-Sec hydrogen price/LCOH-equivalent	$$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$8 in 2030; $$LCOH = \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)}} \quad \text{or} \quad \frac{C_{tot}^{(\alpha)}}{M_{H_2}^{(\alpha)} \times HHV / 1{,}000},$$9 in 2050
Offshore wind power hub	Offshore electrolysis, hydrogen-driven, AEL	minimum $HHV = 39.4$0
UK wind-powered pathways	Configuration-dependent plant LCOH	$HHV = 39.4$1–$HHV = 39.4$2
Corsica 2050 compromise solution	Multi-objective supply-chain design	$HHV = 39.4$3

The German HRS values come from a system study in which nodal electricity prices dominate hydrogen cost and co-optimization lowers average LCOH by $HHV = 39.4$4 relative to the less integrated case (1908.10119). The European endogenous-learning values describe a continent-scale transition in which green hydrogen displaces grey hydrogen under rapid renewable and electrolyzer scale-up (Zeyen et al., 2022). The offshore wind-hub result corresponds to offshore electrolysis in hydrogen-driven mode with alkaline electrolysis and is reported as competitive with current grey hydrogen costs of $HHV = 39.4$5–$HHV = 39.4$6 and at the upper end of blue hydrogen competitiveness at $HHV = 39.4$7–$HHV = 39.4$8 (Singlitico et al., 2021). The UK values show that no modeled case falls below $HHV = 39.4$9, and that the best-performing configuration is the behind-the-meter electrolyser-first, no-back-up case at $\mathrm{MWh}_{\mathrm{H2,HHV}}$0 (Andoni et al., 29 Aug 2025). The Corsica case reports $\mathrm{MWh}_{\mathrm{H2,HHV}}$1 for the equal-weight compromise solution and $\mathrm{MWh}_{\mathrm{H2,HHV}}$2 for a more emissions-favorable alternative (Mai et al., 26 Nov 2025).

Competitiveness against fossil hydrogen remains context-dependent. Under stable natural gas and coal prices, fossil hydrogen is reported at $\mathrm{MWh}_{\mathrm{H2,HHV}}$3, with CCS adding up to about $\mathrm{MWh}_{\mathrm{H2,HHV}}$4 for alkaline, $\mathrm{MWh}_{\mathrm{H2,HHV}}$5 for SOEC, projecting $\mathrm{MWh}_{\mathrm{H2,HHV}}$6, and $\mathrm{MWh}_{\mathrm{H2,HHV}}$7 (Ramadan et al., 2022). By contrast, the IRA study shows that policy-adjusted net LCOH can make green and blue hydrogen cost-competitive with gray hydrogen in the United States even when unsubsidized competitiveness is absent (Cheng et al., 2023).

Several recurrent interpretive issues follow from these results. First, low LCOH does not by itself imply low lifecycle emissions. The U.S. state-level study shows that electrolysis carbon intensity varies strongly with grid mix, and that low-cost states and low-carbon states do not perfectly coincide (Ramadan et al., 2022). Second, low LCOH does not necessarily imply compliance with current “green hydrogen” designation rules. In the grid-connected hybrid renewable plant study, only part of the produced hydrogen qualifies as green under current EU rules even though its emissions remain well below conventional alternatives, and significant CO$\mathrm{MWh}_{\mathrm{H2,HHV}}$8 reductions can be achieved with relatively small increases in LCOH (Farah et al., 2024). Third, seemingly small modeling choices—full foresight, degradation treatment, discount-rate assumptions, or the inclusion of policy credits—can shift reported hydrogen costs by amounts comparable to the cost differences between competing technologies.

Taken together, the literature portrays LCOH not as a universal scalar attached to a technology, but as a boundary-dependent economic functional. Its value emerges from the interaction of electricity procurement, utilization, financing, degradation, infrastructure design, and policy architecture; hence the same electrolyzer can appear noncompetitive, marginally competitive, or strongly competitive depending on whether the analysis is unsubsidized or subsidy-adjusted, plant-level or system-integrated, static or endogenous-learning, and degradation-free or degradation-aware.