Not because the technology fails or the demand disappears. Because most of what gets counted as "pipeline" was never a project in the first place — and the filters that separate the two are getting sharper, not softer.
This is our view, not a published statistic. We think half is generous. Here is the evidence we are reading it from, including the parts that cut the other way.
Wood Mackenzie put Europe's utility-scale storage pipeline at more than 130 GW across 3,000+ projects in 37 countries in February 2026, against roughly 19 GW operational. Taken at face value, that is a conversion rate around 15%.
Take it at face value carefully, though. LCP Delta's EMMES data puts European electrochemical storage at 48.7 GW installed at end-2025 — a very different denominator, reflecting different scope. Anyone quoting a single European storage capacity number without saying which basis they are on is not being careful. We are flagging it because it cuts against our own thesis: on the higher figure, conversion looks less catastrophic.
The EU picture is genuinely strong on its own terms: 27.1 GWh installed in 2025, up 45%, taking the cumulative fleet to 77.3 GWh — roughly ten times the 2021 level. The problem is what sits behind the growth: the 2030 requirement is around 750 GWh. Another tenfold, in four years.
This is the heart of it. Grid connection queues do not count projects. They count applications — and a developer with one site and four possible configurations can hold four positions.
Great Britain has done the arithmetic in the open. NESO's connections reform removed or reclassified 153 GW of battery storage from the prioritised queue. Of the BESS that remains, 83 GW holds Gate 2 offers and 33 GW is slated to connect before 2030 — against an original all-technology queue of around 722 GW. NESO has described the battery queue as roughly three times the capacity implied by Clean Power 2030 targets.
That is a regulator saying, on the record, that most of the storage queue is not going to be connected on the timescale its developers assumed. Every market with a queue this shape will eventually do the same thing. Germany already started: first-come-first-served was replaced on 1 April 2026 with a maturity-based scoring system weighing site control, technical maturity, financial capability and co-location.
Covered above. The binding constraint in most European markets is not capital and not equipment — it is a connection date a lender will underwrite.
EASE reports that permitting for large projects can take up to nine years in some markets, and is lobbying for binding 12- or 24-month deadlines. Nine years is longer than most development equity has patience for, and longer than most option agreements run.
Realised revenues have moved against business plans written in 2022-23. Modo Energy's GB index recorded £47k/MW/year in December 2025, down 45% year-on-year, with the worst single day of that month annualising at £15k/MW/year. German arbitrage spreads compressed from around €140/MWh in summer 2025 to €79/MWh in Q4 2025 as capacity scaled. Cannibalisation is not a forecast risk any more; it is in the data.
The spread between markets is now the whole story. Reported unlevered IRRs put Germany and the Netherlands near WACC breakeven on merchant exposure, while the UK, Italy and Poland reach 12-17% on contracted structures. Projects in the first group do not fail because they are bad. They fail because there is a better version of the same project one border away.
A thesis worth holding should survive its counter-evidence. Ours has some.
Read together, these say something specific: the failure is not economic and not technological. It is allocative. There is more capital and cheaper equipment chasing a fixed number of viable connection points, in a subset of markets where the revenue structure supports debt. That is a filter, and filters have a pass rate.
One honest distinction. A project stuck behind a 2033 connection date has not failed — it has been deferred, and deferral kills projects for reasons that have nothing to do with their merits: option agreements lapse, development funding runs out, sponsors reallocate, equipment quotes expire. Modo applies a working assumption of 20-30% attrition even to capacity that has already won a contract in Poland. Attrition among projects that have won nothing is necessarily higher.
The projects we see reach financial close have most of the following. Very few of the ones that stall have more than two.
None of that is exotic. It is simply the list an investment committee works through, assembled before the meeting rather than during it.
The headline claim is AIP's assessment, not a published third-party statistic; no consultancy publishes a European storage pipeline failure rate, and we are not attributing one. The supporting figures are as publicly reported and current to August 2026, cited to source below, and in two places they conflict with each other on scope — we have said where. This is market commentary, not legal, regulatory, tax or investment advice, and is not an offer or invitation in respect of any investment.
Send the six lines below. A partner comes back within two working days with a straight answer on whether it is financeable now, financeable later, or not financeable — and what would have to change.
Email deals@aip-partners.com with:
We are paid on success, so telling you a project will not fund costs us nothing and saves you a year. If that is the answer, you will get it in the first email rather than after three months of process.
Munich and London. Success-fee only, since 2010. Response within two working days.