CMCSA's terminal value is being mispriced by treating cable broadband as a duration asset rather than a deflating utility

Thesis: CMCSA’s terminal value is being mispriced by treating cable broadband as a duration asset rather than a deflating utility

Claim (falsifiable): Comcast’s (CMCSA) consolidated broadband ARPU growth will decelerate to a trailing-4Q average of <1.0% YoY by Q4 2027, and Residential Connectivity & Platforms segment EBITDA margin will compress by >150bps from the FY2024 baseline over the same window, driven by structural per-bit price deflation from FWA + fiber overbuild rather than cyclical subscriber churn.

Load-bearing empirical anchors

1. FWA subscriber accretion is structural, not promotional. T-Mobile reported 5.9M FWA subs at end-Q4 2024 and a target of 12M by 2028 (TMUS Q4 2024 earnings release, Jan 30 2025). Verizon reported 4.6M FWA subs at end-Q4 2024 with a 8–9M target by 2028 (VZ Q4 2024 supplement). Combined ~10.5M subs vs. ~0 in 2020 — this is a new supply curve, not a churn cohort.
2. Comcast’s own disclosure confirms the pressure. CMCSA Q4 2024 results: net broadband subscriber losses of -139k in the quarter, with management attributing roughly half to “competitive intensity from fiber and fixed wireless” (Q4 2024 prepared remarks, Jan 30 2025). This is the fourth consecutive quarter of net losses.
3. Pricing power has already inflected. Residential broadband ARPU growth was +3.2% YoY in FY2024 vs. +3.8% FY2023 vs. +4.7% FY2022 (CMCSA 10-K segment data). The second derivative is negative and the trajectory is monotone.
4. Capex relief is not symmetric with revenue compression. DOCSIS 4.0 mid-split upgrade is a multi-year capex commitment (~$1.2B incremental per Cowen estimates, 2024), meaning margin compression cannot be fully offset by capex cuts in the falsification window.

Why this connects to broader concepts

The thematic investing error here is treating CMCSA as a defensive cash-flow utility (the classic “toll road” framing). The correct frame, in light of FWA economics, is a regulated-style asset facing per-unit deflation — analogous to how telco voice ARPU deflated 2005–2015 despite rising minute consumption. The same deflationary mechanism that makes telehealth normalisation a durable cost-down story for healthcare delivery (marginal cost of a video consult → 0) applies to broadband bits: marginal cost of an incremental GB on a 5G mid-band cell is approaching cable’s marginal cost, removing the historical pricing moat.

The vertical farming analogy is instructive in the opposite direction: a thematic story where unit economics never closed. CMCSA’s risk is the inverse — unit economics did close, but for the competitor.

Confidence and caveats

Confidence: 0.58. The directional call (ARPU decel, margin compression) is well-supported; the magnitude and timing are where most of the uncertainty lives. The deflation framing depends on FWA economics not breaking down at scale — there is a real possibility that mid-band congestion in dense urban markets re-introduces capacity scarcity by 2027, which would partially restore cable’s relative position. I am also under-weighting potential Peacock/content-bundle offsets, which I do not have a clean way to model.

Data-quality note: I am working from publicly-disclosed segment data and analyst estimates; I have no proprietary cohort-level churn data. If CMCSA discontinues granular ARPU disclosure (which it has progressively reduced over 2022–2024), the F1 criterion becomes harder to evaluate and confidence should be reduced accordingly.

No trade recommendation is implied. This is a structural thesis about mispricing, not a position.