Global hydrogen demand passed 100 million tonnes in 2025. The figure itself is not in dispute. What is disputed, quietly but with real financial consequences, is the unit in which that demand should be discussed. The molecule has been bought, sold and consumed by mass for decades. Refineries purchase feedstock by the tonne. Fertiliser plants contract for ammonia production that depends on hydrogen input measured in kilograms. Steelmakers working with direct-reduced-iron technology assess their hydrogen needs per tonne of iron produced. The commodity enters an industrial process as a chemical reagent, not as an abstract quantity of energy.
The arithmetic of conversion
Nothing prevents anyone from converting hydrogen volumes into energy units. The molecule carries a lower heating value of approximately 33.33 kilowatt-hours per kilogram. Multiply by a thousand and you have megawatt-hours. The arithmetic is sound. The trouble begins when that conversion is treated as analytically neutral, as though choosing to express hydrogen in MWh carries no interpretive weight.
Electricity is generated, traded and consumed in kilowatt-hours because energy is the product. A household buys electricity to run appliances. A factory buys it to power motors. The unit matches the use. Hydrogen, by contrast, is today overwhelmingly manufactured, transported and consumed as an industrial molecule. Expressing it in MWh can make it look like a direct competitor to electricity and gas before it has actually secured a place in those energy markets.
What the EEX HYDRIX benchmark reveals
The European Energy Exchange (EEX) makes the purpose of this conversion unusually explicit. Its HYDRIX methodology allows hydrogen prices to be quoted in euros per kilogram or euros per megawatt-hour, then converts the benchmark to €/MWh so that it can be set alongside electricity and gas prices. The exchange is building price transparency around a developing commodity. That is a legitimate commercial objective. But the framing is also an analytical choice, not merely a change of denominator.
Put electricity at €70 per MWh, natural gas at €40 per MWh and hydrogen at €200 per MWh in the same table. The presentation invites a direct comparison: three energy commodities at different price points, with hydrogen appearing expensive but plausibly substitutable. One megawatt-hour of electricity, however, is already in the form required by motors, electronics, induction heaters and heat pumps. A nominal megawatt-hour of hydrogen represents only chemical heating value. For most energy services the hydrogen must still pass through additional equipment and incur conversion losses. For the industrial uses that dominate actual demand, the heating value may be secondary to the chemical role of the atoms themselves.
The European Energy Exchange is not wrong to build a benchmark in energy terms. Traders need a reference price. Regulators need something to track. But the unit conversion does more than facilitate comparison. It implies a market that has yet to materialise.
Why tonnes describe the market that actually exists
The International Energy Agency has tracked hydrogen demand for years, and the composition has shifted remarkably little. The 100 million tonnes consumed in 2025 went overwhelmingly into refining, ammonia production, methanol synthesis and other chemical processes. These are not speculative future markets. They are established industrial supply chains with known offtakers, known volumes and known contract structures.
An ammonia plant does not buy megawatt-hours of hydrogen. It buys a chemical feedstock by mass because the number of hydrogen atoms entering the Haber-Bosch reaction determines the volume of ammonia leaving it. A refinery has an annual hydrogen requirement tied to its throughput of crude oil and the sulphur content of its feedstock. These are engineering quantities. Recasting them as energy quantities obscures what is actually being purchased and why.
The distinction matters because policy targets and corporate announcements routinely cite hydrogen demand in terawatt-hours. The European Commission's own hydrogen strategy has used energy units to project future demand. So have national strategies from Germany, the Netherlands and Spain. The effect is to make hydrogen look like an energy commodity on a trajectory comparable to electricity or gas, when the underlying demand is still almost entirely chemical.
Applications are not markets
This leads to a larger confusion. A hydrogen application is not automatically a hydrogen market. Space heating belongs to the heat market, where hydrogen competes with heat pumps, district heating, biomass and improved insulation. Hydrogen-fired power generation belongs to electricity, capacity and balancing markets, where it competes with batteries, demand-side response and gas with carbon capture. Grid storage belongs to flexibility and reliability markets, where it competes with pumped hydro, lithium-ion batteries and interconnectors.
Hydrogen can compete as an input to any of these services. But new demand materialises only when customers choose it over the alternatives on delivered cost, emissions, infrastructure and performance. Starting with a modelled 200 terawatt-hours of hydrogen for heating reverses the causal chain: it assumes the application first, then calculates the molecule requirement afterwards. The more rigorous approach is to begin with the service the customer is actually buying, whether that is useful heat, mobility, electricity or grid reliability, and test whether hydrogen wins on the merits.
European gas distribution companies have been among the loudest advocates for hydrogen in heating, partly because their pipelines could deliver it and partly because the alternative, electrification, would make their infrastructure redundant. Their advocacy is understandable but not disinterested. The question for policymakers is whether hydrogen heating wins on total system cost and emissions performance, not whether the gas industry has a compelling reason to promote it.
What investors should measure instead
The same discipline applies to investment. Announcing several gigawatts of electrolyser capacity says remarkably little about a business. Electrolyser capacity is an input metric. It tells you how much hydrogen could be produced if the plant ran at full load, if the electricity price was viable, if there was a buyer for the output and if the logistics of delivery were solved. None of those conditions is automatic.
What matters more: tonnes actually sold, firm offtake agreements with creditworthy counterparties, utilisation rates over time, delivered prices that compete with alternatives, and evidence that the buyer genuinely requires the molecule rather than merely signing a letter of intent to satisfy a policy requirement. Announcements and targets show activity. Operating plants, repeat purchases and high utilisation show that a market exists.
The wave of electrolyser announcements across Europe in recent years has been striking. Germany alone has seen proposals for double-digit gigawatts of capacity. Yet actual operating capacity remains a fraction of what has been announced. The gap between the two figures is itself a signal: permitting, electricity procurement, offtake negotiations and infrastructure take longer than press releases suggest.
The denominator problem in policy and statistics
The HYDRIX framing has a parallel in energy statistics. The United States Energy Information Administration includes hydrogen in its primary energy accounting, which means that hydrogen produced from natural gas is counted as an energy input, and any future hydrogen produced from electricity would be counted again. The result is a statistical picture that can double-count energy flows and inflate the apparent contribution of hydrogen to the energy system.
European statistical frameworks are not immune. Eurostat's energy statistics are still developing their treatment of hydrogen, and there is a risk that the same denominator confusion takes hold: counting hydrogen in energy units without distinguishing between chemical feedstock demand and genuinely new energy demand. The International Energy Agency's tracking of hydrogen demand similarly relies on tonnage for existing uses but projects future demand in terawatt-hours, making the two hard to compare on consistent terms.
The practical consequence is that policy documents can present hydrogen as a larger share of the future energy system than the underlying demand analysis supports. A projection of 200 terawatt-hours sounds substantial. Expressed as roughly 6 million tonnes of hydrogen, it becomes easier to ask whether those tonnes have credible buyers and whether those buyers have alternatives.
Organisations
European Energy Exchange
