Summary
- CityFibre Infrastructure Holdings Limited's audited accounts print 4.4 million premises passed, 4.1 million premises available to order and 518,200 consumer connections in one table at 31 December 2024, and print no take-up percentage inside that table.
- The group's penetration sentence — "CityFibre has now exceeded 20% penetration across its consumer footprint" — divides 848,000 consumer connections by a 4.15 million figure disclosed in footnote 2, which is smaller than both premises totals in the same release; on the release's own 4.5 million RFS figure the ratio is 18.8 per cent.
- The same audited accounts describe the lenders' test as "a capex to debt ratio and a debt to connections ratio". The covenant currency is connections. No document read states the covenant level, the headroom, or the connection count at which that test is met.
Two companies, and only one of them prints a network
The entity matters before any number does. The directory subject is CityFibre Limited, company number 09759465, incorporated on 2 September 2015 at 15 Bedford Street, London WC2E 9HE, and known as NEWINCCO 1376 LIMITED for its first ten weeks (Companies House filing history). It files full accounts on time: the year to 31 December 2022 filed on 17 July 2023, 2023 on 11 October 2024, 2024 on 6 October 2025, and 2025 on 6 October 2026 in a 44-page document.
What those accounts contain is a holding company. Turnover of £113,695,000, an operating loss of £130,886,000, a loss before tax of £448,871,000, net liabilities of £1,339,805,000, cash of £23,496,000 and an average of no employees at all, with four wholly owned subsidiaries underneath it (extract of the parent-only accounts). There is no network table in it, because a parent-only return has no network to table.
The consolidated picture sits one company over. CityFibre Infrastructure Holdings Limited, 08772997, filed group accounts to 31 December 2024 on 7 October 2025 at 83 pages and to 31 December 2025 on 7 October 2026 at 85 pages (Companies House filing history). CityFibre Infrastructure Holdings Limited is the entity whose published document carries the consolidated network KPI table and the group finance lines; CityFibre Limited is the directory subject, a parent-only filer whose own accounts carry no network KPI table. Every reach and connection figure below comes from the first, and the June 2027 covenant warning quoted later comes from the second.
The table prints the inputs and skips the ratio
The report and consolidated financial statements for the year ended 31 December 2024 devote one table to the numbers that matter: revenue £133.9 million, adjusted EBITDA £5.7 million against a negative £54.5 million, capital expenditure £552.0 million against £1,037.4 million, RFS premises additions of 0.9 million, cumulative RFS premises of 4.1 million against 3.2 million, cumulative premises passed of 4.4 million against 3.6 million, consumer connections of 518,200 against 337,000, and right-first-time installation at 95.0 per cent (FY2024 group accounts).
No penetration row appears in it. The term is defined in a footnote — "Penetration rate being the percentage of premises connected to the CityFibre network out of the available RFS premises" — and then used in a cohort sentence: penetration above 40 per cent in the most mature cohorts such as Milton Keynes and Stirling, with early-life penetration nearly doubling between 2021 and 2024. Two definitions sit beside those lines and they are not interchangeable.
Premises passed means "the property has fibre allocation and a fibre route available to the boundary of the premises, but may not yet be connected"; RFS premises means "the property is available to place orders and make installation on and therefore be connected".
Everything needed for the network ratio is in that table. The directors simply do not print it. My arithmetic on their own figures: 518,200 consumer connections over 4,100,000 RFS premises is 12.6 per cent; over 4,400,000 premises passed it is 11.8 per cent. The 300,000 premises between the two totals are built to the boundary and not yet orderable — also my subtraction, from their table.
Then the denominator moves
A year later the company moved the ratio into prose and moved the base with it. The January 2026 trading update, whose table carries the footnote "Management accounts, pending final audit", printed revenue of £170 million, consumer revenue of £114 million, adjusted EBITDA of £29 million, consumer connections of 848,000 cumulative after 330,000 additions, premises passed of 4.7 million and premises RFS of 4.5 million (FY2025 trading update).
In the narrative it says: "CityFibre has now exceeded 20% penetration across its consumer footprint". Footnote 2 supplies the base: "CityFibre's RFS consumer footprint = 4.15m". That figure is 350,000 below the 4.5 million RFS total printed in the same release and 550,000 below its premises passed. My division, on each of the company's own numbers: 848,000 over 4.15 million is 20.4 per cent; over 4.5 million it is 18.8 per cent; over 4.7 million it is 18.0 per cent. thinkbroadband, tracking independently, put RFS at 4.33 million against the company's 4.5 million, which would give 19.6 per cent (thinkbroadband).
The release also prints no interest line, no debt quantum, no covenant level and no headroom. It is a document about customers that happens to be published by a company whose constraint is debt.
What the money is actually keyed to
Here is the sentence that makes the rest of it matter. In the FY2024 accounts, the directors describe the covenants they modelled as "a capex to debt ratio and a debt to connections ratio". Not a take-up ratio. Not a penetration percentage. A debt-to-connections ratio, and a capex-to-debt ratio beside it.
The two quantities therefore move on different clocks. Capital expenditure turns into premises on a construction schedule the company controls, and it did so at £552.0 million in 2024 while adjusted EBITDA sat at £5.7 million. A connection arrives when a household inside a passed area orders service, which the company does not control. The audited table prints both inputs; the covenant contract, on the directors' own description, runs on the second one.
The financing itself is visible. Interest-bearing loans and borrowings rose from £3,092 million to £3,780 million during 2024, against £3.9 billion of committed facilities entered into in June 2022 and fully drawn at the year end, with floating exposure "linked to SONIA" and covenants coming into effect in later years. Finance costs were £279 million, interest costs £292 million, and interest-rate swap settlements gave back £62 million.
The FY2024 going concern expected further funding "in the region of £1.5 billion in the near term, with an expectation of a future capital raise in late 2026"; at the 28 May 2025 signing that money was not secured, and £163 million of shareholder equity sat as cash at Topco.
What the following year's filings say arrives through a third-party extract and secondary reporting rather than the filed PDF. The extract of the FY2025 group accounts gives turnover of £169,576,000, an operating loss of £207,236,000, a loss before tax of £545,406,000, net assets of £1,046,024,000 after the equity injection, and cash of £139,170,000 (group accounts extract). The extract of the parent-only set carries the sharpest sentence in the whole record: "The Group's consolidated financial statements include a material uncertainty related to going concern, arising from the limited headroom on its financial covenant requirements under its debt facilities as at June 2027." (parent accounts extract). The group extract words the same theme as "limited forecast covenant headroom and sensitivity to reasonably possible downside movements". Bloomberg Tax reported the auditor's flag on that June 2027 headroom on 7 October 2026, in a paywalled piece read here only as a search snippet (Bloomberg Tax), and The Telegraph reported on 11 October 2026 that finance costs ballooned to almost £400 million, more than double £170 million of revenue, that borrowing costs pushed the pre-tax loss to £545 million from £360 million, and that the company is seeking to cut a £3.8 billion debt pile in discussions with lenders (The Telegraph). Agreed terms appear in no document read here.
The instance a stranger can check
One named reach figure beside one named connection figure, same document, same date: premises passed of 4.7 million and cumulative consumer connections of 848,000, both in the January 2026 trading update dated 22 January 2026, with the company's own penetration sentence and its 4.15 million footnote base attached. That is the whole mechanism in one paragraph, and it is auditable by anybody with the release.
By 30 September 2026 the position had moved again: the company's 6 October 2026 release, which states that the accounts confirm the January trading update, puts cumulative connections at 1,108,000 against premises passed of 4.8 million and RFS of 4.6 million, all labelled unaudited management accounts (6 October 2026 release). My arithmetic there is 24.1 per cent on RFS and 23.1 per cent on premises passed. A release describing what the accounts confirm cannot establish what the KPI table prints, and the nine-month column is unaudited.
What the record does not say
The 85-page FY2025 group accounts were not obtained from the material read on 19 November 2026; the register metadata and the extracts were, the PDF body was not. Nothing here states what that document's KPI table prints, what its covenant note says, or that it omits a take-up percentage. No document read states a covenant level, a headroom figure, or the connection count at which the debt-to-connections test is satisfied. Consumer customer connections count consumer lines, not whole-network connections, so every ratio above is a consumer ratio and no group-wide take-up figure is asserted.
The 4.15 million is a footnote base, not total RFS and not premises passed.
The directory entry for CityFibre Limited records the parent-only filer, which is the company that files and the company that does not print a network. The number the builder controls is the one that gets printed. The number the customer controls is the one that gets a footnote.
Member Briefing
Read the full trend briefing
Members read the full briefing with the source notes behind every finding.
Strategic Circle
A daily briefing in your language, built from changes we have verified.
Join for USD 20/monthCancel anytime.
Members also get the full research library, a daily briefing on the networks they follow and the private forum.






