Last Updated on August 7, 2026 by Karl Thompson
This article is about why the road outside your house was dug up — possibly three times, possibly by three different companies, possibly not at all. Britain’s full-fibre network is the largest civil engineering project of the past decade and almost none of it was built by the state. It was built by pension funds, sovereign wealth funds and private equity, who bought into the idea that a trench full of glass is a financial asset. This article examines what happens to a country’s infrastructure when it is built to yield a return, and what happens when the return does not arrive.
This is part of How Modern Society Works, a series on the hidden systems that shape everyday life. Each article takes something ordinary and works outwards to the system that produced it.
Why Was My Street Dug Up Three Times?
In a mid-sized English town — Peterborough, say, or Swindon, or Bolton — a residential street may have been excavated three times in five years. Openreach came first, or perhaps CityFibre did, or Netomnia, or brsk. Each dug its own trench, laid its own duct, pulled its own cable and reinstated the tarmac slightly differently from the last. Residents complained about the noise, the parking and the state of the pavement. Then it happened again.
Meanwhile, on a lane in Herefordshire or Cumbria or Argyll, nobody has come at all. The householders there have been waiting since roughly 2019, are still waiting, and have watched at least one company promise to arrive and then withdraw.
These two facts are usually reported separately, as an urban nuisance and a rural grievance. They are the same fact. They are what happens when the decision about where to lay national infrastructure is made by asking where the return is best.
Both are also, in the strict sense, sociological facts: nobody intended either. No planner allocated three networks to one street and none to another. The pattern emerged from many separate investment decisions, each rational, which together produced an allocation of national infrastructure that nobody would have chosen and no one is responsible for.
How Does Fibre Optic Broadband Work?
Before the money, the material — briefly, because the physics is elegant and the economics is the point.
Optical fibre is drawn from ultra-pure silica glass into a strand about the thickness of a human hair. At its centre is a core; around it a layer of glass called cladding with slightly different optical properties. That difference means light striking the boundary at a shallow enough angle is reflected back inwards rather than escaping. The light therefore bounces its way along the fibre, following it round corners and under oceans, in a process called total internal reflection.
Miniature semiconductor lasers switch on and off billions of times a second, turning information into pulses of light travelling at roughly two-thirds the speed of light in a vacuum. And because a single fibre can carry many wavelengths — many colours — simultaneously, each acting as an independent channel, one strand can carry a staggering volume of traffic at once. The cable in the ground is not really a wire. It is a bundle of parallel motorways made of light.
The manufactured cable is cheap. Everything else is not.
Building a fibre network means excavating roads, threading cable through Victorian ducts, erecting or climbing poles, negotiating wayleaves with landowners, avoiding gas mains, managing traffic and reinstating surfaces. The overwhelming majority of the cost is civil engineering rather than the glass itself. Britain’s digital revolution has depended on excavators, traffic management and men in hi-vis at least as much as on software engineers.
This has one consequence that determines everything else in the chapter. The cost of connecting a premises depends almost entirely on how close it is to the next one. A single trench along a terraced street reaches a hundred homes. The same trench in a valley reaches four. The technology is identical; the economics are not remotely comparable.
Geography, in other words, was never abolished. It was converted into a spreadsheet.
Who Owns the UK’s Fibre Network?
Most people buy broadband from a company that does not own the cable entering their house.
Britain’s fixed network is layered. At the bottom sit the infrastructure owners, who hold the physical assets: ducts, poles, exchanges, fibre. Above them sit network operators, and above them the retail internet service providers — Sky, TalkTalk, Vodafone, Zen — who sell to households. One company can occupy several layers, or one.
Openreach, legally separated from BT Group but owned by it, sits at the base of most of this. It inherited something no competitor could replicate: the physical routes. Ducts dug for the telephone system decades ago, poles erected across rural Britain, exchange buildings in every town. The hardest part of building a fibre network is not the fibre; it is the right of way, and Openreach already had it.
Ofcom’s response was Physical Infrastructure Access, a regulatory requirement obliging Openreach to let competitors run their own fibre through its existing ducts and along its poles at regulated prices. This is a genuinely important intervention and it is worth understanding what it did: it converted a natural monopoly asset into a shared platform, and in doing so made it possible for new entrants to build a national network without digging a national network.
The result was one of the more remarkable episodes in recent British industrial history. At its peak, Britain had more than a hundred alternative network operators — “altnets” — building full-fibre networks: CityFibre, Netomnia, Hyperoptic, Gigaclear, Community Fibre, G.Network, Wildanet, Voneus, nexfibre and dozens of others. Over £40 billion of private investment went into the ground.
Where did £40 billion come from? Not from telecoms companies. It came from finance.
CityFibre is backed by Antin Infrastructure Partners, Goldman Sachs, Mubadala Investment Company — the sovereign wealth fund of Abu Dhabi — and Interogo Holding, the Liechtenstein-based foundation associated with IKEA. Hyperoptic is owned by KKR. Gigaclear is backed by Infracapital. nexfibre is a joint venture involving InfraVia, Liberty Global and Telefónica.
These are not technology investors. They are infrastructure funds, and they were not buying broadband. They were buying an asset class.
Why a Trench Became a Financial Asset
To understand why global capital wanted to dig up Swindon, it helps to look at what infrastructure funds are actually seeking.
An infrastructure asset, in the financial sense, has a characteristic shape: enormous capital cost up front, very long life, high barriers to entry, and — once built — a stream of small, recurring, contractually stable payments from a large number of customers who have limited alternatives. Roads, airports, water companies, energy networks and now fibre all share that shape. Pension funds in particular need long-duration, inflation-resistant income to match long-duration liabilities. A network that costs a fortune to build, lasts forty years and collects a monthly fee from every household on the street is, on paper, exactly that.
David Harvey’s account of capital switching describes this general movement precisely. When surplus capital cannot find sufficient returns in ordinary production, it switches into the built environment — into long-lived, fixed, spatially embedded assets. The built environment absorbs capital that has nowhere else profitable to go. Harvey’s further argument is the one that matters here: this switch does not resolve the underlying problem so much as relocate it, and it characteristically ends in over-accumulation — too much capital committed to fixed assets that cannot generate the returns assumed — followed by devaluation.
Harvey’s account is influential and it is also contested. Economic geographers who have gone looking for the switching dynamic in aggregate investment data have had difficulty demonstrating it, which is worth stating rather than glossing. What the framework offers here is not proof that capital moved from production into fibre, but a description of a pattern — surplus capital seeking long-lived fixed assets, over-committing, and then devaluing — that the altnet story fits closely.
Between roughly 2018 and 2023, in an era of very low interest rates, an enormous quantity of capital switched into British fibre. The thesis was straightforward: build past as many premises as possible, as fast as possible, sign up customers, and either hold the asset for its yield or sell it to someone who will.
Two things then happened. Interest rates rose sharply, which made the debt behind these projects far more expensive to service. And take-up disappointed.
How Many People Actually Use Full Fibre?
The coverage figures for Britain’s fibre rollout are genuinely impressive, and worth stating accurately.
As of July 2025, Ofcom recorded full fibre available to 78% of UK residential premises — 23.7 million homes — an increase of nine percentage points in a single year. Gigabit-capable coverage, which includes Virgin Media’s cable network alongside full fibre, reached 87%, rising to 89% (27.1 million premises) by January 2026. Northern Ireland leads at around 95% full fibre; Scotland trails at around 71%, reflecting its islands and glens.
Then there is the other number.
Of the premises that can get full fibre, only 47% had actually taken it. For most of the rollout period the figure was lower still — around 42% in July 2025.
More than half of the households with a full-fibre network passing their front door have not connected to it.
This is one of the most sociologically interesting facts in the whole series, and it exposes something the language of “coverage” hides. Availability is not access, and access is not use. Households do not upgrade because a trench appeared. They upgrade when a contract ends, when a price is right, when someone in the house needs it, when they understand what it is. Many are on adequate superfast connections and see no reason to pay more. Some cannot afford to. Some do not know the option exists.
There is an unexpected wrinkle: in England, take-up among rural premises runs materially higher than urban — around 54% against 39%. Where the alternative is genuinely poor, people connect. Where the existing copper or cable service is tolerable, they don’t. Demand for infrastructure turns out to be shaped by what people are already living with, not by what is technically available.
For investors, this gap between homes passed and homes connected was catastrophic, because the entire financial model rested on converting the first into the second.
What Happened to the UK’s Altnets?
By 2024, cumulative accounting losses across the UK altnet sector had reached around £1.5 billion. Only one operator, Hyperoptic, had reached EBITDA-positive status by conventional measures. Networks had been built past millions of premises that were not paying for them, funded by debt that had become expensive.
What followed is a fairly orderly financial crisis, of a kind rarely covered outside the trade press.
G.Network, which had dug up much of central London, entered administration in January 2026 before being reorganised under new ownership by FitzWalter Capital. Gigaclear, the rural specialist, completed a recapitalisation in April 2026 that delivered an estimated 40% haircut to its lenders — meaning the people who financed rural fibre lost close to half their money. CityFibre, the largest altnet, carries roughly £3.7 billion in net debt against total liabilities reported at £5.1 billion, has raised new financing worth £2.26 billion, and has run repeated rounds of redundancies through 2026. Netomnia cut around fifty roles in February 2026 while awaiting a takeover.
That takeover is the consolidation endgame. In early 2026, nexfibre — the joint venture behind Virgin Media O2’s network expansion — agreed to acquire Substantial Group, the parent of Netomnia, YouFibre and brsk, for approximately £2 billion, in what would be by far the largest altnet transaction to date. CityFibre objected publicly, arguing the deal risked re-establishing a duopoly of BT and Virgin Media O2. The Competition and Markets Authority’s scrutiny is expected to determine the shape of the market.
Step back from the corporate detail and the pattern is Harvey’s, almost to the letter. Capital switched into the built environment at scale. Over-accumulation followed. Devaluation is now under way — through administration, through lender haircuts, through consolidation at prices well below what was invested. And the physical asset, notably, does not disappear. The fibre stays in the ground. It simply changes hands at a lower price.
That last point deserves emphasis, because it complicates any simple story of failure. Britain got a fibre network. It was substantially paid for by investors who will not recover what they put in. Whether that constitutes a policy triumph — a national asset built at the expense of global capital — or a policy failure — a chaotic, duplicative build that has now chilled further investment — is a genuine argument with serious people on both sides.
What is not in dispute is who ended up with the network. As the altnets contract, Openreach — the incumbent, with the ducts, the poles and the balance sheet — is targeting 30 million premises by 2030. The competitive market that private capital was supposed to create may end up strengthening the position it was supposed to challenge.
Why Is Rural Broadband Still So Slow?
Return now to the second street, the one nobody came to.
Commercial fibre investment stops where the return does. That boundary is not a technical limit but a financial one, and it falls at different places depending on the cost of capital, so it has moved outwards and then back inwards over the past decade. Somewhere beyond it are roughly the final tenth to fifth of British premises: farms, hamlets, islands, the ends of long lanes.
Project Gigabit, the government’s £5 billion programme, exists precisely to reach them — the state as builder of last resort, arriving where the market declines to. Its progress is instructive.
According to Building Digital UK’s own published contract data, by July 2026 the programme had connected 287,510 rural premises out of 838,490 contracted — around 34%. Roughly 551,000 homes and businesses under contract were still waiting. At the recent build rate, completing them would take approximately another three years.
The reasons are as revealing as the numbers. Several altnets that won Project Gigabit contracts subsequently walked away from them — FullFibre Limited exiting the Peak District and West Herefordshire, Wildanet withdrawing from Cornwall, alongside Voneus and Freedom Fibre — citing delivery costs higher than anticipated. Openreach has picked up many of these contracts as fallback supplier, but the new contracts were awarded late and their builds are barely begun. The largest single contract in the programme, covering the rest of Scotland, stood at around 7% complete.
So the sequence, for a household at the end of a long lane, ran roughly: the commercial market decided you were unprofitable; the state stepped in and contracted a private firm to reach you; that firm discovered you were unprofitable too and withdrew; the incumbent inherited the obligation and has not yet arrived.
This is the sharpest available illustration of a general principle. Where infrastructure is delivered through markets, the state’s role becomes residual — cleaning up the segments that private capital has priced out. And residual provision inherits the same underlying economics that made those segments unattractive in the first place. Subsidy changes who pays; it does not change the fact that reaching four houses costs what reaching a hundred costs.
Infrastructure With a Yield Attached
Fibre optic cable is a good candidate for the most consequential British infrastructure of the past twenty years. It underpins remote work, cloud computing, streaming, digital public services, online banking and — as later chapters will argue — the viability of an economy that has increasingly assumed connectivity in the way it once assumed electricity.
It was built almost entirely with private money, on the premise that it would function as a financial asset. That premise organised everything: which streets were dug, in what order, how many times, and which were left. The map of Britain’s fibre rollout is not a map of need. It is a map of expected return, and where those two diverged, return won.
This is what financialisation means in practice. It is not a synonym for privatisation, and it is not simply about who owns things. It is about infrastructure being evaluated as an asset — assessed for yield, duration and exit value — and therefore built, or not built, according to criteria that have nothing to do with the social purpose the infrastructure serves.
The criteria are not hidden or sinister. They are published in investor presentations. But they answer a different question from the one a country might ask. An infrastructure fund asks: where will this generate a stable return over twenty-five years? A society asks: who needs to be connected? Most of the time in urban Britain those questions produced the same answer, which is why the rollout succeeded as far as it did. Where they diverged, only one of them was being asked by anyone with a budget.
And now that the returns have disappointed, the capital is leaving — through administration, through haircuts, through consolidation — while the trenches remain. Infrastructure outlives the financial thesis that produced it. That is true of Victorian sewers, of the railway mania of the 1840s, and of a hundred fibre companies that dug up Britain between 2018 and 2023.
The cable under the pavement will carry light for decades. The reason it is there at all will have been forgotten within one.
Data centres in west London were allocated electricity by a queue. Fibre was allocated by expected yield. In both cases the mechanism doing the allocating was never presented to anybody as a decision about national priorities, because in neither case did it feel like one to the people operating it.
The next chapter concerns what happens after the cable reaches the building — and finds the same pattern arriving at its logical conclusion. Fibre investors at least wanted to own something. The organisations at the other end of the cable have increasingly stopped owning anything at all.
References
| Claim | Source | URL | Why it supports the claim |
|---|---|---|---|
| Full fibre available to 78% of UK residential premises (23.7m) as of July 2025; +9 percentage points year on year; gigabit-capable 87% (26.4m) | Ofcom, Connected Nations UK Report 2025 | https://www.ofcom.org.uk/siteassets/resources/documents/research-and-data/multi-sector/infrastructure-research/connected-nations-2025/connected-nations-uk-report-2025.pdf?v=407947 | Section 2, residential coverage figures |
| Gigabit-capable coverage 89% (27.1m) by January 2026; full-fibre take-up 47% of premises with access | Ofcom, Connected Nations update: Spring 2026 | https://www.ofcom.org.uk/phones-and-broadband/coverage-and-speeds/connected-nations-update-spring-2026 | Interim update headline figures |
| Rural take-up 54% vs urban 39% in England; Northern Ireland 95% full fibre, Scotland 71% | Ofcom, Connected Nations England Report 2025 and nations reporting | https://www.ofcom.org.uk/phones-and-broadband/coverage-and-speeds/nations-report-2025 | Take-up and nation-level coverage tables |
| Physical Infrastructure Access: duct and pole access remedies | Ofcom, duct and pole access | https://www.ofcom.org.uk/phones-and-broadband/telecoms-infrastructure/duct-pole-access-remedies | Sets out the regulatory remedy and its purpose |
| Over 100 fibre operators at peak; over £40bn of private investment; altnet cumulative accounting losses £1.5bn in 2024; only Hyperoptic EBITDA-positive | Bratby Law, UK fibre consolidation analysis | https://bratby.law/uk-fibre-consolidation-transactions/ | Sector-level financial summary with figures |
| G.Network entered administration January 2026, reorganised under FitzWalter Capital; Gigaclear recapitalisation April 2026 with ~40% lender haircut | Capacity, reporting Assembly Research analysis | https://capacityglobal.com/news/nexfibre-calls-for-uk-consolidation/ | Names both events and the haircut estimate |
| CityFibre net debt c.£3.7bn; backed by Antin, Goldman Sachs, Mubadala, Interogo; further redundancies proposed 2026 | ISPreview | https://www.ispreview.co.uk/index.php/2026/07/broadband-altnet-cityfibre-notifies-200-uk-staff-of-possible-future-redundancy.html | States ownership, debt position and redundancy round |
| CityFibre total liabilities reported at £5.1bn | ISPreview, reporting Assembly Research | https://www.ispreview.co.uk/index.php/2026/06/assembly-research-examines-uk-full-fibre-market-and-nexfibres-netomnia-deal.html | Cites the liabilities figure |
| nexfibre agreed to acquire Substantial Group (Netomnia, YouFibre, brsk) for c.£2bn; CityFibre objected citing duopoly risk | Bratby Law, altnet M&A wave analysis | https://bratby.law/uk-altnet-consolidation-fibre-ma-wave-2026/ | Details the transaction, parties and objection |
| Openreach targeting 30 million premises by 2030 | Capacity, reporting Assembly Research | https://capacityglobal.com/news/nexfibre-calls-for-uk-consolidation/ | States the Openreach target in context of altnet stall |
| Project Gigabit: 287,510 rural premises built of 838,490 contracted as at July 2026 | Building Digital UK, July 2026: Premises contracted and built under Project Gigabit contracts | https://www.gov.uk/government/publications/premises-contracted-and-built-project-gigabit-contracts/july-2026-premises-contracted-and-built-under-project-gigabit-contracts | Official monthly contract-level progress data |
| FullFibre, Wildanet, Voneus and Freedom Fibre exited Project Gigabit contracts; Openreach acting as fallback supplier; Scotland contract c.7% complete | Tech Times analysis of BDUK July 2026 data | https://www.techtimes.com/articles/321144/20260721/rural-uk-still-waiting-two-thirds-gigabit-contracts-remain-unbuilt.htm | Names the withdrawing operators and the fallback arrangement |
| Project Gigabit budget £5bn; aim of c.99% gigabit coverage by 2032 | ISPreview, Project Gigabit progress coverage | https://www.ispreview.co.uk/index.php/2026/07/july-2026-contract-progress-of-uk-project-gigabit-broadband-rollout.html | States programme budget and target |
Note on sourcing. The financial condition of individual altnets is reported by trade press and analyst firms rather than in a single official publication, because most of these companies are privately held. Where an official source exists — Ofcom’s coverage data, BDUK’s contract data — it is cited in preference. The Openreach civil-engineering cost share, which appears in many industry discussions as “60–80% of build cost”, has been stated qualitatively here rather than numerically, because I could not locate a primary source for the specific range.
Reading
- David Harvey, The Limits to Capital (Blackwell, 1982) — capital switching into the built environment. See also the empirical critiques of the switching thesis in economic geography
- David Harvey, The Urban Experience (1989) — the secondary circuit and over-accumulation
- Susan Leigh Star, ‘The Ethnography of Infrastructure’ (1999)
- Nicholas Blomley, Rights of Passage (2011) — on the pavement as contested legal and social space
How Modern Society Works
← Previous: What Are Data Centres? · Series hub · Next: What Is Cloud Computing? →