More than 14,500 companies are in the window

The maturity wall has been forecast for three years, mostly from the top down: scheme-lending vintages, leveraged-loan samples, survey sentiment. This counts it from the bottom up, company by company, from the public register. We scored 83,957 UK lower-mid-market companies against Companies House charge data. 26,383 carry an outstanding registered charge, held by a named incumbent, whose age implies an estimated refinancing window. 14,675 are inside that window today. For most of the companies on the wall it has already opened; it is not a 2027 event still on the horizon.

What we counted, and how

Companies House publishes every security interest a UK company grants: who holds it, when it was created, what it covers. It does not publish facility amounts or maturity dates, so we estimate them. Each outstanding charge gets a tenor prior keyed on what it looks like — a clearing-bank or challenger debenture reads as a four-year facility give or take eighteen months, a single fixed or floating charge as three-year asset finance, a debt-fund charge as a six-year unitranche — and the window is the creation date plus that prior. Rolling facilities, invoice finance and ABL are excluded: they carry no refinancing clock. Rent deposits, pension-scheme and landlord charges are excluded: they are not facilities. One company, one primary window.

The limits are worth stating plainly. Every window in this dataset is an estimate, typically three years wide, and none is a reported maturity. A charge proves security was granted, not that the facility is still outstanding; repaid facilities routinely stay on the register because satisfaction filings are optional. And the 83,957 companies are a prioritised sample of a much larger universe, so every count here is a floor, and every share describes this sample rather than the UK economy. That is the confidence statement for the whole piece: the quarter a given company lands in is soft; the size of the cohorts, the origination vintages and the concentration of who holds the paper are hard. Everything that follows leans on the hard part.

The shape of the wall

Because the windows are years wide, the honest measure is not how many facilities mature in a given quarter but how much paper sits inside its refinancing zone at any given moment. On that view the wall reads as a plateau rather than a spike: between 15,519 and 16,808 estimated windows are open in every single quarter from the start of 2026 through mid-2027. The peak quarter, 2027Q2 at 16,808, is barely higher than the quarter we are in. The count only tapers meaningfully from late 2027, down to 9,282 open windows by the end of 2028.

The shape decides how the wall behaves. A spike is a scheduling problem; a plateau is a capacity problem, because the same lender credit teams, debt advisers and legal capacity have to clear an elevated volume for eight consecutive quarters. A borrower who assumes the crowd thins out next year is misreading it: the pressure holds through mid-2027 before it eases.

Where the wall came from

Strip out the estimation layer entirely and the wall is still there, because it is a vintage story, and vintages are hard register dates. Charge creation among the dated cohort concentrates overwhelmingly in 2020 through 2024: 4,373 charges created in 2020, 4,494 in 2021, 4,837 in 2022, 5,360 in 2023 and 5,585 in 2024. Facilities papered together age into their refinancing windows together. The pandemic-era borrowing, the 2021 recovery lending and the repricing wave of 2022 to 2024 are now, on customary three-to-five-year senior terms, one long queue arriving at the same set of doors.

Split those vintages by who holds the paper and the composition barely moves from year to year: at this end of the market the register shows a bank-held book throughout. Clearing banks carry the mass of every single vintage, from 1,042 of the 2019 cohort to 2,919 of 2024’s. Challenger and specialist banks are a modest, slowly growing band. Recognisable debt funds never exceed 59 charges in any year — a reminder that the private-credit story so visible in the upper mid-market has barely reached the register at £3–15m. The rest is trustee-held paper, where a real facility is implied but its lender is not disclosed, and corporate holders the classifier does not recognise.

Three positions on the wall

The 26,383 dated companies split three ways against today’s date, and each position carries a different job. 5,684 are pre-window: their estimated windows have not opened yet. This is the strongest position on the wall and the least used one — everything about a refinancing is cheaper here, from time to fix the credit story to the option to simply wait if terms are wrong. The work at this stage is preparation, not transaction.

14,675 are in-window now. For a performing business this is a question of timing rather than solvency, and timing is precisely what the plateau squeezes. With fifteen-thousand-plus windows open every quarter, lender attention is the scarce resource, and it goes to the borrower who arrives prepared and early inside their window rather than late in it.

And 6,024 are past their estimated window end, at a median of 0.6 years ago. This cohort is two populations the register cannot tell apart: companies running past their facility’s natural life — extended, amended, or drifting on evergreen terms that suit the incumbent — and companies that have already refinanced or repaid, with the old charge sitting unsatisfied on the file because nobody filed the form. The aggregate overstates the truly overdue by the unfiled-satisfaction rate, which nobody can measure from public data. But the first population is real, it is large, and a borrower in it has usually stopped choosing their terms without noticing.

Who holds the paper

Among the 14,675 in-window companies, 13,750 carry a named incumbent — 1,667 distinct charge-holding entities — but the distribution is anything but flat. Folded to banking groups, five names hold 52.5% of the primary charges: HSBC 17.0%, NatWest Group 12.7%, Barclays 11.1%, Lloyds Banking Group 9.3% and Santander UK 2.4%. Clearing banks as a category hold 54.9%. A further 925 charges sit with bare security trustees, where an institutional facility is implied but the lender behind it is not disclosed, and 4,068 sit with corporate holders our classifier does not recognise. Those unknowns are counted in the denominator, so the bank shares above are, if anything, understated.

Both of the obvious readings of that concentration are wrong. It does not show that banks are the wrong home for this paper: for most of these borrowers a bank facility was, and often still is, the cheapest and simplest answer, and plenty of these refinancings should end exactly where they started. Nor is the concentration harmless. When five credit desks face half the wall at once, each one is triaging, and its pricing on any given renewal reflects its book, its sector appetite that quarter and its capacity — none of which the borrower can see. The practical consequence is the same whichever way the individual decision goes: an in-window borrower facing a concentrated incumbent base has more to gain from a real competitive process than in any normal year, even, and sometimes especially, when the best outcome is the incumbent matching a better term sheet.

The date to work backwards from

None of this is a forecast. It is a count of paper that already exists, held by lenders already named on a public register, aging on terms the market itself set between 2020 and 2024. For a borrower, the single actionable fact in 83,957 rows of data is your own charge’s creation date — it is on your Companies House file today — and the customary tenor of what you signed. Add them together and you have the date the market will expect you to move. The 14,675 companies inside their window are not a market statistic. Each one has a specific date, and the register has been showing it all along.

Aggregate register data only. Method, priors and every “what this is not” caveat are set out in full in the dataset’s methodology note, and the vintage counts are published as an open specimen dataset for anyone who wants to check them against the same file. The full report, typeset — PDF →, alongside the rest of the example work. The standing reference page, kept current as the count moves: the UK refinancing wall, 2026–2028.