Chapter 2

The Broadband Base

Charter's broadband customer count peaked at 30.59 million at the end of 2023 and has fallen every year since, to 29.56 million at March 2026 [1] [2]. Until this year, higher rates more than covered the lost units. In the first quarter of 2026 they did not: rate and mix added 9 million dollars to residential Internet revenue against 87 million lost to volume [3]. That crossing is what this chapter examines.

Internet is 43% of revenue — $23.8 billion of $54.8 billion in 2025 [4] — and it is the largest single input to the cash flow that services the debt.

The shape of the decline

Sources: FY2022 Form 10-K [5], FY2023 Form 10-K [6], FY2024 Form 10-K [7] and FY2025 Form 10-K [8]; Q1 2026 earnings release [9]. Residential customers only; the totals quoted in the text include small business.

The 2024 loss of 510,000 residential customers has a partial one-off explanation: the FCC's Affordable Connectivity Program ended in the second quarter of 2024, and Charter attributes a share of that year's disconnects to it [10]. The 2025 loss of 393,000 does not, and the first quarter of 2026 was worse than the first quarter of 2025 — 120,000 total Internet customers lost against 59,000 [11].

Penetration of the footprint fell from 56.7% at December 2024 to 54.0% at March 2026 [12] [13], but two-thirds of that move is the denominator. Estimated passings grew from 56.86 million to 58.40 million during 2025 [14]; held at the 2024 footprint, the 2025 relationship count of 31.85 million would represent 56.0% penetration rather than the reported 54.5% — a decline of 0.7 points, not 2.2. Charter is building into its own penetration ratio, which is a choice, not a symptom.

Two disclosure changes sit inside this series and should be handled with care. In the fourth quarter of 2024 Charter restated its December 2023 passings from 56.99 million to 55.32 million [15]. Then, in the fourth quarter of 2025, it revised customer relationship statistics "to include all mobile customers, including mobile-only customers" [16]. Prior periods were restated, so the series is internally consistent, but the effect on the level is material: December 2024 relationships were reported as 31.47 million in January 2025 [17] and as 32.21 million a year later [18], a difference of 741,000 customers who take a phone line and no wire. The headline relationship count is no longer a proxy for broadband households. Internet customers are the clean series, and that is the series used throughout this chapter.

Where the customers went

Three national wireless carriers now sell home broadband over their cell networks, and their combined book grew by about four million connections in 2025 — roughly three times the broadband customers the four listed cable operators lost between them.

Sources: T-Mobile FY2025 Form 10-K, customer metrics [19] and FY2024 Form 10-K [20]; Verizon FY2025 Form 10-K [21] and FY2024 Form 10-K [22]; AT&T FY2025 Form 10-K [23]; Comcast FY2025 Form 10-K [24]; Altice USA FY2025 Form 10-K [25]; Cable One FY2025 Form 10-K [26]; Charter Q4 2025 earnings release [27]. T-Mobile's figure combines postpaid and prepaid 5G broadband. Comcast's change is reported net losses; its year-end level also reflects roughly 124,000 business customers added through the Nitel acquisition and excluded from net losses.

The arithmetic does not prove causation — household formation, copper displacement and fibre all sit inside the same gap — but it establishes scale. Fixed wireless is no longer a rounding error against a 30-million-customer base, and its fastest-growing entrant is the newest: AT&T added 875,000 Internet Air connections in 2025, more than doubling its book [28]. Management's own reading matches. Asked in April 2025 whether fibre overbuilders were splitting his markets, Chris Winfrey said the fibre effect "has been consistent and steady" and that the real drivers were "mobile substitution and the introduction of a new low-end competitor with cell phone Internet" [29]. A year later the framing was unchanged, and sharper: the problem is "a top-of-funnel issue", with churn "at historical lows" and yield at the point of sale "as strong as ever" [30].

Fibre overlap is the second front, and Charter's own disclosure of it is harder to read than it looks. In the 2021 Form 10-K, AT&T, Frontier and Verizon competed in approximately 34%, 9% and 5% of operating areas at speeds of at least 25 Mbps [31]; by 2023 that was 35%, 11% and 6% on the same definition [32]. The 2024 filing raised the threshold to the FCC's 100 Mbps standard, and the same three names fell to 25%, 9% and 6% [33]. In 2025 Frontier disappears from the sentence entirely and the figures are AT&T 27% and Verizon 16% [34]. That jump is a merger, not a construction programme: Verizon completed its acquisition of Frontier on 20 January 2026 [35], and the 2024 figures of 9% plus 6% land almost exactly on the 2025 figure of 16%. The change in the level of overbuild is small; the change is that two competitors became one, with one balance sheet and one converged offer.

No chart is offered for that series, because the 25 Mbps and 100 Mbps definitions are not the same measurement and plotting them together would suggest a fall in overbuild that did not happen.

Price against volume

Charter publishes its own decomposition of residential Internet revenue into rate-and-mix and volume. For three years it told a consistent story: rate covered the units.

Sources: derived from Charter's own revenue bridges — FY2023 Form 10-K [36], FY2024 Form 10-K [37], FY2025 Form 10-K [38] and Q1 2026 Form 10-Q [39]. Each contribution is expressed as a percentage of the prior period's residential Internet revenue. Q1 2026 is a quarterly year-over-year comparison, the others annual.

In 2025 rate and mix added $785 million while volume subtracted $380 million, for net growth of $405 million on a $23.36 billion base [40] — the strongest rate contribution of the four periods shown. In the first quarter of 2026 the rate line collapsed to $9 million against $87 million of volume loss, and residential Internet revenue fell 1.3% year over year [41] [42]. At the 2025 run rate, each 1% of residential Internet revenue is about $238 million a year.

Three qualifications belong in the same breath. First, one quarter is one quarter. Second, Charter attributes even that $9 million partly to "a favourable change in bundled revenue allocation" [43] — allocation between the Internet, mobile and video lines moves revenue around without changing what the household pays, which cuts both ways on the FY2025 figure too. Third, and most substantively, the household is still spending more: mobile service revenue grew 15.1% in the quarter, and connectivity revenue — Internet plus mobile together — still rose 0.9% [44]. What has changed is the rate of that offset: connectivity revenue grew 4.1% in 2025 [45] and 0.9% in the first quarter of 2026.

Where the price competition is visible is in what Charter now pays to acquire a household. In the first quarter of 2026 it launched a guarantee of $1,000 of first-year savings to customers who take Spectrum Internet and move two or more mobile lines from Verizon, AT&T or T-Mobile [46]. Headline broadband pricing holds; the discount is routed through the bundle.

What the capital is buying

Roughly half of Charter's capital budget goes to the footprint rather than to the existing customer.

Sources: FY2023 Form 10-K, capital expenditure table [47]; FY2025 Form 10-K, capital expenditure table [48]. Upgrade/rebuild is the category that contains the network evolution initiative; all other combines customer premise equipment, scalable infrastructure and support capital.

The network evolution initiative — symmetrical and multi-gigabit speeds across the whole footprint — sits inside upgrade/rebuild, which was $1.94 billion in 2025 [49]. Half the network is due to be upgraded by the end of 2026, with completion in 2027 [50] [51]. The subsidised rural build is the larger line: $2.20 billion in 2025 alone and $7.7 billion since 2022, activating about 1.3 million passings towards a target of more than 1.7 million, against more than $2 billion of government support awarded [52] [53].

The rural build works, and it is masking what happens elsewhere. Charter added 39,000, 47,000, 52,000 and 46,000 customer relationships inside the subsidised rural footprint through the four quarters of 2025 [54] [55] [56] [57] — 184,000 in total, against a company-wide fall of 368,000 [58]. The established footprint therefore lost roughly 552,000 relationships in 2025, half again the headline. That gap will not close by itself: rural passings growth is guided to 450,000 in 2026, described by management as the last large build year [59].

What the evidence supports

Charter's competitive position is narrow rather than wide, and the numbers that establish it are relative, not absolute. Against the listed cable operators it is the best of a losing set: Charter's broadband base fell 1.3% in 2025, Comcast's 2.2%, Altice USA's 4.7% and Cable One's residential data base 5.8% [60] [61] [62] [63]. It also penetrates its footprint far more deeply — 54.5% of passings at the end of 2025 against Comcast's 47.6% of homes and businesses passed [64] [65], though the two measures are defined by each company and are not strictly comparable.

What that advantage does not extend to, on this evidence, is pricing power. A moat should be visible in the ability to raise price faster than units decline; in the first quarter of 2026 it was not. The read here is that Charter's capital programme is aimed at the part of the problem that is already working. Retention is not the constraint — management describes churn at historical lows and better still where a mobile or video product is attached [66] [67] — and a faster network is a retention and pricing asset. Acquisition is the constraint, and a symmetrical multi-gigabit upgrade does not obviously restore a household's willingness to consider Charter when its phone already carries a home broadband offer.

The strongest fact against that read is that Charter's losses are concentrated in gross additions rather than disconnects, which is exactly the pattern one would expect from a shrinking pool of movers rather than from a product deficit: management points to a muted housing environment, slow household formation and low move rates alongside the new competition [68] [69]. If that is right, the base stabilises when housing turnover normalises, without Charter doing anything differently, and the network upgrade arrives in time to monetise it. That claim cannot be tested from this corpus — Charter publishes no gross-add or churn figures, only its characterisation of them, and no share data for its fibre-overlap markets beyond the assertion that its share there remains above the competition [70].

Two observable things would change the read. Residential Internet revenue returning to growth with a rate contribution back above 2% would show the pricing engine restarting and make the first quarter of 2026 a blip. Conversely, a second and third quarter in which rate contributes near zero while volume subtracts 1.5% would establish that the household's broadband dollar is now falling in both directions at once — and at that point the relevant arithmetic is not the customer count but the $4.9 billion run-rate annual cash interest that a 43%-of-revenue line has to keep covering [71].