Every techno-economic model of a hydrogen project has a storage line, and in most of them it holds a single number in euros per kilogram. That number is doing more work than any other cell in the model, because three separate things are hiding inside it: whether the site has the geology at all, how often the storage turns over, and whether it can be built before the plant needs it.
None of the three is a cost assumption. They are answers to questions about a specific place.
The rock decides, and it is not evenly distributed
Figure 1
Where the salt is
Modelled technical potential for hydrogen storage in onshore salt caverns, by country. Europe's onshore total is 23.2 PWh.
The largest onshore potential in Europe by a wide margin.
Second, and not usually the country named in a European storage conversation.
Third. An order of magnitude below Germany.
Published as 510 TWh — 2% of the overall potential, despite widespread bedded-salt deposits.
Germany, Poland and Spain account for 77% of the total onshore potential between them. France is in the chart to make a specific point: it has widespread bedded salt and still lands at 2% of the overall potential, because thickness, depth and geometry decide, not the presence of salt on a map.
Source: Caglayan et al., Technical Potential of Salt Caverns for Hydrogen Storage in Europe · Data as of 2019-10-16
Show the data ▸Hide the data ▾
| Country | Onshore potential | As published |
|---|---|---|
| Germany | 9.45 PWh | 9.45 PWh |
| Poland | 7.24 PWh | 7.24 PWh |
| Spain | 1.26 PWh | 1.26 PWh |
| France | 0.51 PWh | 510 TWh, 2% of the overall potential |
| Europe, onshore total | 23.2 PWh | 23.2 PWh |
| Europe, onshore and offshore | 84.8 PWh | 84.8 PWh, 42% of it Germany |
The reference study for European salt cavern potential models cavern geometry, salt thickness, depth window and pressure limits across the continent, and reports the totals plainly:[1]
the overall technical storage potential across Europe is estimated at 84.8 PWhH2, 42% of which belongs to Germany
Restrict that to onshore and it falls to 23.2 PWh.[1] Within that:[1]
The highest potential is observed in Germany (9.45 PWh), followed by Poland and Spain (7.24 PWh and 1.26 PWh). These three countries account for 77% of the total onshore storage potential.
The instructive entry is France, and it is instructive precisely because the naive expectation is wrong:[1]
the onshore storage potential is estimated at 510 TWh, which is only 2% of the overall storage potential
France has widespread bedded salt. It is on the geological map. And it converts to 2% of the potential, because a cavern needs the salt to be thick enough, at the right depth, in a body large enough for a cavern of a workable size with the required spacing between caverns. "There is salt in this country" and "this country can store hydrogen underground" are different claims, and the gap between them is where a project plan goes wrong.
The cost is a function of cycling, not of geology
Figure 2
The same cavern, at two hold times
Levelised cost of storing hydrogen in a salt cavern, from one model, for a 15-day and a 120-day hold.
Executive summary of the same report states $0.15/kg for this case.
Executive summary of the same report states $1.2/kg for this case. Table ES-1 gives a range of 0.6–1.2 for a 2–4 month duration.
Same cavern, same model, same report. What changes between the two bars is how long each kilogram sits underground before it is sold. A storage cost quoted without a hold time or a cycling frequency is not a cost — it is a number waiting for an assumption. The report's own executive summary gives $1.2/kg and $0.15/kg for these same two cases; Table 9 gives the figures shown here.
Source: UC Davis Institute of Transportation Studies, UCD-ITS-RR-24-17, Table 9 · Data as of 2024-02-27
Show the data ▸Hide the data ▾
| Hold time | Table 9 | Executive summary |
|---|---|---|
| 15 days | $0.235/kgH2 | $0.15/kg |
| 120 days | $1.59/kgH2 | $1.2/kg |
Here is the finding that undoes the single-number storage line. In one model of one salt cavern, the levelised cost of storing a kilogram of hydrogen is $0.235 on a 15-day hold and $1.59 on a 120-day hold.[2] Nothing about the rock changed. What changed is how many kilograms the same capital moves in a year.
This is the same structure as a capacity factor, and it deserves the same discipline. Seasonal storage — filling in summer, emptying in winter — is one turn a year, and every euro of capital is amortised over that single turn. Weekly arbitrage storage is fifty turns, and the same capital is spread across fifty times the throughput.
So a model that carries "storage: €X per kilogram" without stating the cycling frequency has not made an assumption about cost. It has made an assumption about the operating pattern of the plant, invisibly, and put a price on it.
And the alternative to a cavern is not slightly more expensive
The same report gives capital costs for compressed gas tanks at hydrogen refuelling stations: about $700/kgH₂ of capacity for steel with a liner and about $650/kgH₂ for a steel-concrete composite, at 1,500 kg and 350 atm.[2] Against that, preparing a 500-tonne salt cavern is put at about $18M, or $36/kgH₂ of capacity.[2]
Those two figures are not on the same boundary and this page does not adjust them onto one — a refuelling-station tank package and a cavern preparation cost are answering different questions, and forcing them into a single ratio would be exactly the error this blog spends its time on. But the order of magnitude survives any reasonable reconciliation, and it is the whole strategic point.
A project without access to a cavern is not paying a premium for storage. It is in a different cost regime, and the correct response is usually to redesign around needing far less storage — flexible downstream operation, a smaller buffer, a firmer grid connection — rather than to put tanks in the model and carry on.
Nobody has committed to this yet
Figure 3
Announced storage, and committed storage
Of all underground hydrogen storage announced for delivery by 2035 — around 11 TWh, or 325 kt H₂ — the share that has reached final investment decision or is under construction.
Of around 11 TWh (325 kt H₂) announced for 2035.
The IEA states this equivalence directly.
Both figures are the IEA's, stated as percentages in the same sentence. The second is the more useful one for a project: it says that the committed storage covers 2.5% of what the committed production projects will make in a year. The other 97.5% of that output has no committed place to go when it cannot be sold the moment it is made.
Source: IEA, Global Hydrogen Review 2025 — Trade and infrastructure · Data as of 2025-09
Show the data ▸Hide the data ▾
| Measure | Value |
|---|---|
| Underground hydrogen storage announced by 2035 | around 11 TWh (325 kt H₂) |
| Share that has reached FID or is under construction | 5% |
| Equivalent share of annual output from committed production projects | 2.5% |
The IEA states it in a single sentence:[3]
If all announced underground hydrogen storage projects are realised by 2035, around 11 TWh of capacity (325 kt H₂) will be available. However, only 5% has reached FID or is under construction, equivalent to 2.5% of the annual output from committed low-emissions hydrogen projects.
Read the second half twice. It is not saying that 5% of a storage wish-list is real, which would be unremarkable. It is saying that the storage which is actually being built covers 2.5% of what the committed production projects will produce in a year.
Every one of those production projects has a plan for what happens to a kilogram that cannot be sold in the hour it is made. For 97.5% of that output, the plan is not a hole in the ground that anyone has yet decided to pay for.
The timeline does not fit inside a project schedule
The most useful public evidence here is a project that publishes its own calendar. HyStock, at Zuidwending in the Netherlands, states that permits and a spatial integration decision were required from 2022; that Nobian had completely leached the first cavern, A5, by 2023; that exploratory drilling for the future caverns happens in 2025; that construction of the above-ground facility and infrastructure commences in 2028/2030; that debrining then begins by filling the cavern with hydrogen as cushion gas; and that the first cavern will be ready for use around 2031.[4]
Nine years from the start of permitting to first storage service — with the cavern itself physically finished in year two.
That ordering is the finding. The geology was the fast part. What consumed the remaining eight years was permitting, spatial planning, surface infrastructure, connection and the decision to spend the money. Anyone whose schedule assumes that storage can be added once the electrolyser is ordered has the sequence backwards: on this evidence, storage is the long-lead item and the electrolyser is not.
What to do with this before your next site selection
- Ask the geology question before the cost question. Not "what does storage cost" but "is there a cavern-capable salt formation under or near this site, and does anyone have rights to it". If the answer is no, the storage line in the model is fiction regardless of the number in it.
- State a cycling frequency next to every storage cost. Turns per year, or hold time in days. Without it, the number is unfalsifiable and, on the evidence above, can move by a factor of seven.
- Put storage on the critical path, not in the operating cost. If a cavern is part of the plan, its permitting starts before the plant's, not after. HyStock's own schedule is the argument.
- Check the primary table when a storage cost arrives second-hand. A widely cited report disagrees with itself by roughly a third between its summary and its results table. That is not an outlier; it is what happens to any number that gets quoted more often than it gets opened.
- Design the buffer down before pricing it up. Downstream flexibility, firmer offtake profiles and a grid connection all substitute for stored kilograms. They are usually cheaper than tanks, and always cheaper than a cavern that does not exist.
The uncomfortable version of all this is short. For most European sites, there is no cavern, and there will not be one on this project's timeline. That is a constraint to design around, not a cost to estimate — and a model that prices it as a cost has already given the wrong answer.