Comcast and the Arithmetic of a Narrowing Moat
What the business is
A moat can remain visible after the water has begun to drain. Comcast still owns formidable infrastructure, but the direction of travel matters more than the postcard.
The economic core is Connectivity & Platforms, which sells internet access over Comcast’s cable network to roughly 31.4 million homes and businesses. In 2026 Q1, the segment produced $19.96 billion of revenue and $7.9 billion of EBITDA. Around that foundation sit Xfinity Mobile, an MVNO with more than 9 million lines, NBCUniversal’s studio and streaming operations including Peacock, and the theme parks. Epic Universe opened in 2025, while 2026 Q1 park revenue increased 24%. Most cable channels, including USA, CNBC and MSNBC, were separated into Versant in 2025.
Historically, Comcast’s moat came from efficient local scale and cost advantage. Once a cable network had been built, serving another subscriber carried a low marginal cost. Limited local competition supported annual price increases without much customer flight.
That arrangement is weakening. Broadband subscriber losses continued during 2025, and Comcast is moving most customers to a simpler, discounted pricing structure by the end of 2026. Cutting prices to preserve the relationship may be sensible, but it is not evidence of pricing power. A toll bridge becomes less attractive when the owner starts handing out coupons.
Fiber construction is extending into Comcast’s territory, and fiber take-rates in overlapping markets are around 20 percentage points higher than cable’s. T-Mobile and Verizon fixed wireless access also compete for the lower-priced segment. DOCSIS 4.0 is Comcast’s technical response, but cable retains a disadvantage against fiber in upload symmetry.
The mobile bundle may reduce churn, while parks could become a more important earnings pillar. Even so, I see convergence primarily as the cost of defending the moat, not proof that it is widening.
What the numbers say
The price-adjusted valuation is arresting:
- Market cap (price-adjusted, 2026-10-12 close): $70.44 billion
- Owner-earnings yield (price-adjusted): 34.72%
- Share price (2026-10-12 close): $20.68
- 10-year Treasury: 5.24%
The spread between the owner-earnings yield and the Treasury is +29.48 percentage points. Mechanically, the valuation clears the quantitative circle rubric, including its market-cap condition. Cheapness, at least on the supplied screen, is not hiding behind the curtains.
The earning-power bridge deserves more care. The disclosed starting points are 2025 net income of $20 billion and actual free cash flow of $19.2 billion. The thesis then identifies a possible one-time Versant separation benefit in net income, the effect of park capital spending ending, continuing capital expenditure of $11.8 billion, and declining broadband earnings. It applies a no-growth multiple of 10 times after normalization.
What the thesis does not provide is a dollar amount for each adjustment. That means the normalized endpoint and resulting intrinsic-value range cannot be independently reconstructed from the stated bridge. I therefore treat the screen as evidence of a very low price relative to current reported cash generation, not as a precise appraisal of durable earning power. Precision without a reproducible bridge is merely confidence wearing spectacles.
Capital allocation is similarly mixed. Comcast returned $11.7 billion during 2025, including $6.8–7.2 billion of repurchases. Shares outstanding declined 5%, and the dividend increased 6.5% to $1.32 per share. Repurchasing stock at a depressed valuation can be useful if the underlying earnings are durable.
The longer record is less reassuring. Comcast paid $39 billion for Sky in 2018 and recognized an $8.6 billion impairment in 2022. The controlling Class B structure also carries an undilutable 33⅓% of total voting power. Recent discipline is welcome, but accounting tends to remember enthusiasm long after presentations have moved on.
What Wall Street believes vs what I see
The recent headlines are increasingly constructive. They frame the Versant separation as a possible cure for the conglomerate discount, emphasize the dividend, and suggest that a rebound may follow a decade of underperformance. The prevailing mood appears to be shifting from structural decline toward cheap assets with identifiable catalysts.
I understand the appeal. A 34.72% owner-earnings yield against a 5.24% Treasury is an unusually large discrepancy. If present cash generation proves durable, the market’s pessimism will look rather theatrical.
The difficult word is “durable.” Broadband demand itself is not the central uncertainty. The uncertainty is whether Comcast can retain that demand, and at what margin, as fiber and fixed wireless expand. The pricing reset turns an abstract moat debate into an observable test: after the reset, can Comcast resume increasing average revenue per user without worsening customer losses?
I also assign less weight to corporate catalysts when governance and prior capital allocation have destroyed substantial value. A separation can make an organization easier to understand. It does not restore pricing power, improve network economics or guarantee that future cash will be deployed wisely.
The low valuation is real. So is the possibility that the market is discounting continued erosion in broadband economics rather than merely displaying bad manners.
Verdict
Verdict: pass.
The quantitative rubric indicates circle because the 34.72% owner-earnings yield comfortably exceeds the 5.24% Treasury and the $70.44 billion price-adjusted market cap satisfies the rubric’s valuation condition. I still pass because the moat is narrowing, pricing power is under pressure, the capital-allocation record is uneven, and the intrinsic-value bridge is not fully reproducible. I would reconsider if broadband subscriber losses approached zero, pricing increases resumed after the reset, fiber expansion slowed, and disciplined repurchases and asset rationalization continued.
This analysis is AI-generated, educational, and not investment advice. Figures may contain errors or be delayed. Disclaimer