Full Fibre Passes Two-Thirds of the UK and 40% Have Bought It

Openreach’s full fibre network reaches 23.5 million premises, about two-thirds of the United Kingdom. It has cost £15 billion to build.

Some 9.4 million customers have actually taken it. That is a take-up rate of roughly 40% of the premises passed.

On 19 August the company said it would expand its faster XGS-PON tier from around 40,000 premises to more than a million, which raises an obvious question about the six in ten who have not bought the first version yet.

The Gap Between Passing and Selling

Passing a premises and connecting it are entirely different pieces of work, and only one of them earns anything.

Passing means the fibre runs down the street and into the local infrastructure, so a household could order the service. Connecting means someone did. The first is a civil engineering cost the network operator carries in full; the second is the revenue event, and it depends on a consumer deciding to switch.

A 40% conversion rate is not a failure. It is roughly what a network at this stage of rollout would expect, because a large share of the 23.5 million was passed only recently and those households have not yet reached the end of a contract. Take-up climbs for years after a street is built.

But it does explain why the economics of a national fibre build are so unforgiving. The capital goes out at the moment of passing, in full, whether or not anyone orders. The return arrives gradually, household by household, over the following decade.

What £15 Billion Buys

Divided across 23.5 million premises, the build works out at roughly £640 per premises passed. Divided across the 9.4 million actually connected, it is closer to £1,600 per paying line.

Those two numbers frame the whole business. The first is the engineering achievement and the one usually quoted. The second is the figure the investment case has to clear, and it moves only when take-up rises.

The targets extend further. Openreach aims for 25 million premises by the end of 2026, with an ambition of up to 30 million by 2030, described as subject to conditions. Since 23.5 million represents about 66% of the country, the implied total is somewhere near 35.6 million premises, so 30 million would be the large majority of it.

The remaining premises are also the expensive ones. Networks build the dense, cheap areas first, and what is left toward the end of a rollout is rural, scattered and costs multiples per home of what the early urban work did.

The Upgrade Inside the Upgrade

XGS-PON is a second generation of the same fibre. The strand in the ground does not change; the electronics at each end do.

The current network runs mostly on GPON, which offers 2.5Gbps down and 1.24Gbps up as shared capacity across the households on a given fibre split. In practice the fastest consumer products deliver around 1.8Gbps down and 120Mbps up.

XGS-PON raises that to as much as 8.5Gbps symmetric. The trial covers roughly 40,000 premises around Guildford and Woking, expanding to more than a million by the end of August 2027, starting with exchanges in southern England before reaching all four nations. Over five years the company is aiming at around a third of its full fibre footprint. Trials of 100Gbps Cablelink are expected in 2027.

Trevor Linney, Openreach’s Director of Network Technology, framed the base position simply: “Our Full Fibre network already offers Gigabit capable broadband to more than 23.4 million premises.”

Upload Is the Real Change

The download headline is the least interesting part of this.

Going from 1.8Gbps to 8.5Gbps down changes almost nothing for a household, because essentially no domestic activity is constrained by a 1.8Gbps connection. Streaming several 4K films at once uses a fraction of it.

Upload is different. Moving from 120Mbps to 8.5Gbps symmetric is a factor of about seventy, and upload is where consumer connections have always been starved. It is the constraint that bites on video calls with several participants, on cloud backup, on sending large files, and on anything self-hosted.

That makes this a business-user upgrade far more than a consumer one. A small firm running from a residential-grade line, a video production company, an architecture practice moving large models, or anyone whose staff work from home and push data upward rather than pull it down, gets something they can measure. A household streaming television does not.

Which is a reasonable segment to target when six in ten of your existing footprint have not bought the standard product. Selling a premium tier to businesses is a different market from persuading a household to switch.

Full Fibre Is About to Mean Two Different Things

The five-year target is around a third of the full fibre footprint on XGS-PON. Which means roughly two-thirds of it stays on GPON.

Both will be sold as full fibre, because both are. The strand is identical and the marketing term describes the medium rather than the electronics. But one delivers 120Mbps upload and the other delivers up to 8.5Gbps, and a factor of seventy is not a detail.

For a household choosing a streaming package that distinction is invisible. For a business deciding where to put an office, or a person deciding whether they can work from a particular house, it is the whole question, and it will not be answerable from the words on the tariff.

This is a familiar pattern in UK broadband. “Superfast” and “fibre” both spent years covering a wide range of actual performance, and consumers had no reliable way to tell what they were buying without checking the underlying technology. The same ambiguity is now being rebuilt one layer up.

It also creates a geography. XGS-PON starts in southern England exchanges before reaching all four nations, so for a period the faster tier will be available in some places and not others, on a network already marketed nationally under one name.

None of that is a criticism of the engineering, which is a straightforward upgrade path and cheaper than laying anything new. It is a point about what a buyer can actually find out, and the answer for the next few years is: not much, without asking which generation of equipment serves their street.

One ISP Is Currently Selling It

EE is the only internet provider offering XGS-PON packages at the moment.

Openreach is a wholesale network. It does not sell to households; it sells access to retail providers who do. A new tier is therefore worth nothing until several of them build products on it, price it and market it.

One ISP at launch is normal, and EE sharing a corporate parent with Openreach makes it the obvious first mover. The signal to watch is when the second and third arrive, because that is when a wholesale product becomes a market rather than a pilot.

Until then, the addressable demand for 8.5Gbps symmetric is limited by distribution rather than by technology or by the network build.

What Take-Up Has to Do Next

The number that matters over the next two years is not premises passed. It is the 40%.

Every percentage point of take-up on a 23.5 million footprint is 235,000 additional paying lines on infrastructure that has already been paid for, which is close to pure contribution. Nothing else in the business has that operating leverage.

The barriers are mostly not technical. Households sit on contracts, do not know full fibre is available, or cannot see why they should pay more for speed they do not feel short of. That last one is the hard one, and it is the same problem the industry has faced at every previous speed step.

It is also why the upload story deserves more prominence than it gets. A capability people can feel the absence of sells itself in a way that a larger download number does not, and it is a more durable argument than raw speed. That pattern shows up across infrastructure: capacity gets built ahead of the demand that eventually justifies it, and the interval is uncomfortable, much as it is for a construction sector whose order book and output move on different clocks. It also sits alongside the wider finding that the measurable gains from new technology land first with the firms that build it.

About The Author

Will Lisil

Founder of MW3.biz, building technology that empowers creators and democratising access to digital tools.

See author's posts