Comcast trades at 6.5x earnings and 6.1x EV/FCF on $25B of TTM owner earnings while losing the broadband subscriber war to fiber and fixed wireless. The math is generous; the trajectory is the question.

Plain English

Comcast sells you internet, cable TV, cell service, and runs Universal theme parks. Most of its money comes from internet to homes and businesses. New competitors like fiber-optic companies and T-Mobile's home internet are stealing customers, so the customer count is shrinking — but each remaining customer pays more, and the business still gushes cash. The stock is unusually cheap (about 6 years of profit buys the whole company). The bet: even if customers slowly leave, the cash machine works for years, and the parks plus business-internet keep growing.

Thesis

Comcast is two businesses bolted together: (1) a regulated-like cable connectivity utility (Residential Connectivity & Platforms plus Business Services Connectivity) that throws off most of the cash, and (2) a media-and-experiences conglomerate (Peacock, NBC/Universal, theme parks) that consumes a lot of capital and sometimes earns its keep. After the January 2026 Versant spin-off [filings], the parent is more squarely a connectivity-plus-content business, with the most economically sensitive cable networks (USA, CNBC, Syfy, MS NOW) carved out.

The scorecard frames the setup. ROIC averaged 18.1% over ten years and incremental ROIC over the last five was 54.5% — both reflect a fixed cost base where any incremental dollar of broadband or wireless ARPU drops to the bottom line. FCF conversion of 1.26x means GAAP earnings under-state cash. Net debt is reported at -0.107x EBITDA in the scorecard, but the gross balance sheet shows ~$94.6B of debt against $9.5B of cash; the negative ratio reflects an EBITDA-and-cash framing I'd treat with skepticism — call leverage moderate, not zero. Share count has crept up 5.4% over a decade despite ~$2B/quarter of buybacks, because comp dilution and acquisitions have offset.

The price/IV math is the entire pitch. At $27.19 against scorer IV of $130/$188/$297, the px/IV ratio is 0.14. Even a 50% haircut to base IV ($94) is a 3.5x. Owner earnings of $25B against a market cap near $99B implies a ~25% trailing earnings yield. The only way this isn't cheap is if owner earnings are about to step-function down — which is exactly the bear case the inversion takes seriously.

At $27, you are paid to wait. I want to own this if (a) broadband subscriber declines stay in the 1-2%/yr range while ARPU offsets revenue, (b) wireless and BSC keep compounding at 10-15%, and (c) Epic Universe / parks earn their cost of capital. That is plausible, not certain. Buy below $32, trim above $55.

Moat

Comcast's moat is best understood per-segment, because the conglomerate averages a wide moat in cable connectivity with a narrow-to-none moat in media.

Cost advantage in cable plant (WIDE, narrowing). The hybrid fiber-coax network passes ~63 million homes and businesses. Re-creating that footprint costs tens of billions and a decade of construction; this is a textbook scale-economies moat — fixed costs spread across a dense subscriber base produce a unit-economics gap competitors must out-spend or out-bundle to close. Damodaran's framing on legal/structural advantages applies: local franchise rights and the difficulty of trenching new fiber in built-out neighborhoods give Comcast a quasi-licensed position [2]. The erosion risk is not theoretical: AT&T, Verizon, Frontier, and T-Mobile/Verizon fixed wireless are spending $20B+/yr collectively to overbuild. Q1'26 domestic broadband revenue fell to $6,338M from $6,679M (-5.1% YoY) [filings] — the first time the cable cash cow has reversed at scale. Verdict: still wide on cost-to-serve, but the addressable share is shrinking.

Switching costs (NARROW). Bundles of broadband + mobile + video create real friction — Comcast's wireless line growth (+15% revenue YoY) is largely a retention tool, MVNO-economics over Verizon's network. But broadband is increasingly a commodity: a fiber installer arrives, and the customer leaves in an afternoon. Buffett's preference for Cap Cities/ABC and Coke [1][3] hinged on switching costs that strengthened with use; Comcast's strengthen with bundled discounts that compress ARPU.

Intangibles — content (NARROW for parent post-Versant). The most defensible content (Universal IP, NBC Sports, premier sports rights, theme park brands) stayed with Comcast; Versant got the linear cable brands [filings, Note 6]. Universal/Illumination/DreamWorks IP is real but cyclical, and Peacock is sub-scale vs. Netflix/Disney+. Theme parks have genuine moats — Universal Orlando + Epic Universe (opened 2025) + Hollywood + Beijing + Osaka is a global oligopoly with Disney. Theme Parks revenue grew 24% YoY in Q1'26 [filings]; that segment's moat is genuinely wide.

Network effects (NONE). No two-sided network economics here. Peacock would like to be one but isn't.

Pricing power (NARROW and weakening). Comcast historically raised broadband prices 3-5%/yr; that stopped working in 2023-2024 as fiber overbuilds let competitors hold price. Going forward, ARPU likely flat-to-down on standalone broadband, with growth from wireless attach and BSC.

Competitor stress test ($10B + 5 years). Could T-Mobile spend $10B over five years and meaningfully damage cable broadband? It's already happening — T-Mobile and Verizon added ~5M fixed wireless subs while cable lost net subs. Could a fiber overbuilder kill Comcast in any single market? Yes, locally. Could anyone replicate the entire footprint plus the parks plus Universal IP? No.

Erosion risk. The base case is Comcast loses 5-10% of its broadband subscribers to fiber and fixed-wireless over 5 years, but defends ARPU through wireless bundling and DOCSIS 4.0 speed upgrades. The bear case is share loss accelerates and ARPU breaks.

The Buffett canon emphasizes that brand-and-cost moats only work when management strengthens them [2] — Cap Cities did, Quaker/Snapple didn't. Comcast's DOCSIS 4.0 upgrade and wireless attach are the strengthening playbook; the open question is whether a HFC plant can hold against pure fiber over a 10-year horizon.

Moat verdict: NARROW (was WIDE; cable plant moat is in the late innings; theme parks and BSC are durable wide-moat pieces; standalone-broadband moat is eroding faster than bulls admit).

Management

Capital allocation under Brian Roberts (Chairman/CEO) and the recent Co-CEO structure (Mike Cavanagh) is competent at execution and mediocre at the big swings.

Reinvestment. Capex runs ~$12B/yr against ~$25B owner earnings. The mix has shifted from video set-tops to network upgrades (DOCSIS 4.0, mid-split, distributed access architecture) and to theme parks (Epic Universe, ~$7B over the build). Incremental ROIC of 54.5% suggests the marginal dollar earns its keep — but that figure is flattered by Theme Parks ramp and could reverse if broadband goes ex-growth. I'd grade reinvestment a B: largely defensive on cable, genuinely offensive on parks and wireless.

Acquisitions. Mixed-to-poor record. NBCUniversal (2011/2013) was the deal of a generation in hindsight — bought when GE was a forced seller. Sky (2018, ~$39B at the top of a bidding war with Fox) was a value-destroyer; Sky has been written down multiple times and underperforms in a structurally challenged European pay-TV market. The 2026 Versant spin [filings] is a tacit admission that the cable-network bundle (USA/CNBC/Syfy) was not a long-term holding and bulking up via the Time Warner Cable bid (blocked) and other linear-TV moves added complexity. Net: Roberts overpaid for Sky; the spin partially undoes 25 years of conglomerate-building.

Debt. Carries ~$94.6B of debt (Q1'26 [filings]) at a weighted-average cost in the 4-5% range. Investment-grade, well-laddered, no near-term refinancing wall. This is conservative for a cash-generative utility-like business and reasonable given content cyclicality.

Buybacks. $1.5B repurchased in Q1'26 plus $1.25B dividends [filings]. At current prices ($27, P/E 6.5x, EV/FCF 6.1x, px/IV 0.14), buybacks are extraordinarily accretive — the company is essentially shrinking the share count at a 15%+ earnings yield. The problem is Comcast bought stock at $50-$60 in 2021-2022 and is buying it at $27 now; the average P/IV at purchase has been mediocre, not great. Share count has still risen 5.4% over a decade despite the program, because of comp grants and stock-funded deals. Grade on buybacks alone: B at current prices, C-grade looking back at the cycle.

Dividends. Steady raises, ~3.6% yield at current price. Reasonable.

Communication. Comcast IR is professional, clear in segment disclosure, and forthright about subscriber losses (they don't bury the number). The 2026 segment re-cut to align around Co-CEOs and the Versant carve-out is a genuine attempt at clarity, not opacity. Roberts has been candid that wireless and BSC are the growth pillars and that broadband subs will keep declining. That candor is worth something.

Insider alignment. Roberts' Class B shares give him outsized voting control [filings risk factors], a governance flaw Buffett would dislike. He has skin in the game but cannot be voted out — the Sky deal is the clearest case where shareholder discipline was missing.

The 5-choice scorecard: Reinvest B, Acquire C (Sky overhang), Debt B+, Buybacks B at $27, Dividends B, Communication B+. Net synthesis: a steady operator whose worst decisions came from empire-building and best decisions came from playing defense and prosecuting the cable cost curve.

Capital allocator: B (would be B+ without Sky and without the dual-class voting structure).

Industry

Threat of new entrants — HIGH and rising. The single most important industry change of the last five years: capital markets and the federal BEAD program are funding fiber overbuilds at unprecedented scale. AT&T plans 50M fiber passings; Frontier, Ziply, Brightspeed, and dozens of municipal/co-op builders are layering on. T-Mobile and Verizon turned spare 5G capacity into fixed wireless and added ~5M subs to a service that didn't exist in 2020. Cable's geographic monopoly on high-speed home internet — the most valuable franchise in American telecom for 25 years — is becoming a duopoly or triopoly market by market.

Bargaining power of suppliers — MEDIUM. Programming costs are the headline supplier issue: NFL, NBA, college sports rights inflate at 8-12%/yr while consumer willingness to pay for linear video shrinks. The Q1'26 income statement shows programming and production at $10.9B vs $8.4B prior year (+29%) [filings] — though that includes Epic Universe ramp and rights timing, the trend is clear. On the network-equipment side, suppliers (CommScope, Casa, Vecima, Harmonic) are weak; Comcast has scale leverage. On talent, NBCU/Peacock pays competitive but not crazy rates.

Bargaining power of buyers — INCREASING. Residential broadband used to be take-it-or-leave-it; now consumers have 2-3 viable options in most metros. Cord-cutting has turned video into a buyer's market. BSC (small/medium business) is more durable — switching is sticky and Comcast's hybrid fiber footprint is competitive. Theme parks have pricing power because the experience itself is the moat.

Threat of substitutes — HIGH. Streaming substitutes for linear video; mobile (4G/5G) substitutes for some home broadband at the low end; satellite (Starlink) substitutes in rural markets; YouTube/TikTok substitute for premium video for younger cohorts. The substitution is gradual but relentless.

Industry rivalry — INTENSIFYING. A decade ago, Comcast competed with Charter and DirecTV. Now it competes with Charter, AT&T fiber, Verizon FiOS+5G, T-Mobile fixed wireless, Frontier, dozens of fiber overbuilders, plus Disney/Netflix/Amazon for content. Marketing intensity is up; promotional broadband ARPU has compressed.

Value pool location and trajectory. The historical value pool — residential cable broadband — is shrinking in subscribers but holding in ARPU. New value pools that Comcast plays in: wireless (MVNO economics, +15% revenue YoY), BSC (~$10B/yr revenue and +6% YoY [filings] — high-margin, sticky), theme parks (24% revenue growth Q1'26 with Epic Universe). The dying value pool: linear TV distribution (Versant's now-spun problem); Sky pay-TV in Europe.

Net assessment. This is no longer the wonderful business it was in 2015-2019. It is a fair-to-good business at a great price. Theme parks and BSC are wonderful. Residential broadband is structurally challenged but profitable. Media is fair. Sky is poor.

Industry Verdict: Average (was Good/Excellent five years ago; the structural reset is real and not cyclical).

Inversion

I am now short Comcast at $27. Here is why I will be right.

The single event that kills this. Domestic broadband subscribers move from a 1-2%/yr decline to a 5-7%/yr decline as fiber-to-the-home covers 70%+ of Comcast's footprint by 2030 and fixed wireless becomes good enough for the price-sensitive tail. Once subscriber losses pierce ~5%/yr, ARPU offsets stop working — you cannot raise prices on a customer who is actively shopping the alternative. Domestic broadband revenue then compounds down at -3 to -5%/yr instead of the modest decline visible in Q1'26 [filings]. Wireless and BSC are too small to plug the hole — wireless at ~$1.4B/quarter cannot offset broadband's $6.3B/quarter going negative. Operating leverage runs hard in reverse on a fixed-cost network.

Why the moat is narrower than bulls think. Bulls anchor on Comcast's historical 18% ROIC and assume the franchise endures. The franchise was a regulated quasi-monopoly that depended on (a) no overbuilders, (b) wireless being a different product, and (c) video being bundled with broadband. All three are gone. AT&T, Verizon, T-Mobile, and BEAD-funded fiber co-ops are systematically removing the geographic monopoly. Wireless is now a substitute, not a complement, for the bottom 30% of households. Video has decoupled from broadband entirely — Netflix doesn't care which pipe you use. The cost-to-serve advantage in HFC plant is real but narrowing as fiber capex per passing falls and as cable's necessary upgrade capex rises. DOCSIS 4.0 is buying time, not winning the war. Theme parks are a fine business but represent ~$8B of revenue against $130B+ total — they cannot anchor the equity value.

Why management is worse than it appears. Brian Roberts overpaid by ~$15B for Sky in 2018 in a bidding war he should have walked away from; that capital is gone. The Time Warner Cable pursuit (blocked 2015) would have made Comcast bigger but not better. Peacock has burned ~$10B+ cumulatively and is sub-scale against Netflix and Disney+. The Versant spin is the right move five years late — those linear cable networks were obviously melting in 2020 and management held them anyway. The Class B voting structure means accountability is a one-way conversation. The Co-CEO appointment (Cavanagh) signals succession ambiguity, not clarity. Capital allocation is graded on the easy decisions (debt management, dividends) and gets a pass on the hard ones (Sky, Peacock, footprint M&A).

What bulls are extrapolating that won't hold. Bulls extrapolate (1) ROIIC of 54% from a period that included the lowest cable competition in two decades and the post-COVID broadband bonanza. The next five years will not look like the last five. (2) FCF conversion of 1.26x partly reflected working-capital and depreciation timing; mean reversion to 1.0-1.1x is more likely. (3) Theme parks ramping Epic Universe is a one-time stair-step, not a compoundable trend — parks revenue cannot grow 24%/yr indefinitely. (4) Wireless economics are flattering today because Comcast resells Verizon's network at MVNO rates; if Verizon repricesthat MVNO agreement upward (the contract has commercial reset clauses), wireless margins compress hard. (5) The buyback yield is real but uses cash that should be hoarded for inevitable network capex acceleration as fiber overbuild forces a defensive response.

Valuation trap (multiple compression / regime change). Comcast trades at 6.5x earnings because the market correctly senses that the underlying earnings stream is declining. The bull says 'cheap'; the bear says 'value trap, like AT&T at 7x in 2018, or Intel at 10x in 2020, or any number of telcos and legacy media that compounded down for a decade.' The reverse-DCF question is: at what terminal growth rate does $27 become fair value? At 0% growth and a 10% discount, $25B of owner earnings is worth $250B — far above today's market cap. But if owner earnings decline 3%/yr in perpetuity (a regime, not a cyclical dip), present value collapses to $192B, then $150B, then below today's enterprise value once you net out debt. The IV math the scorer produces ($188B base) implicitly assumes flat-to-modest growth. Replace that with -2 to -3%/yr secular decline and IV halves. Multiple compression toward 4-5x earnings is then justified, putting the stock at $17-$20, not $27.

If I am right, the stock could be worth $15-18 within 3-5 years.

Lollapalooza Bias Check

Biases active in me right now, ranked by force:

Anchoring (high). I am anchored on the scorer's IV range of $130-$297. That range is generated by a model with stated uncertainty (the scorer notes flag maintenance capex as having a >50% spread). Anchoring on $188 base IV makes a $27 price look absurdly cheap — but if maintenance capex is structurally higher than the model assumes (because cable has to invest more to hold against fiber), true IV could be 30-50% lower. I should weight my range toward $90-$130, not $130-$200.

Authority/canon bias (medium). Buffett's letters [1][3] talk about wonderful businesses with durable moats — Cap Cities, Coke, Wells Fargo. The temptation is to pattern-match Comcast onto 'Cap Cities with cable instead of broadcast' and decide it deserves a Buffett-multiple eventually. Buffett actually owned Charter (sold it). Pattern-matching to canon icons that didn't have a structurally declining subscriber base is exactly the mistake.

Confirmation bias (medium-high). A 6.5x P/E and 0.14 px/IV ratio is the kind of headline that makes value investors feel righteous. I am instinctively building the bull case and looking for reasons the bear is overdone. The honest move is to invert hard, which I tried to do in the inversion section.

Recency bias (medium). Q1'26 broadband revenue was -5.1% — that's one quarter, after Versant carve-outs and DTC reclassifications [filings]. Treating one quarter as the new run-rate would be foolish; ignoring it as noise would also be foolish. The honest interpretation is somewhere in the middle.

Deprival super-reaction (low-medium). Comcast was a $50+ stock five years ago. Anchoring on that price creates a feeling that current investors are getting away with something at $27. But the business that supported $50 is partly gone; the price is right for what remains, not for what was.

Commitment / consistency (low). No prior position; clean slate.

Social proof (low). Most value investors I respect either own this or owned it recently (Dodge & Cox, Berkshire historically owned Charter not Comcast). The crowd effect is mildly bullish but not overwhelming.

Incentive bias (medium, on management's side). Roberts is incentivized to keep the empire intact and to time spins for tax efficiency. Versant is good news on incentive alignment but doesn't reset the dual-class governance flaw.

Net read: I should mark down the IV range, take the inversion seriously, and refuse to let a low headline multiple seduce me into ignoring secular decline. The action that follows is to size small and require a margin of safety below current price.

10-Year Outlook

Will the same fundamental business model exist in 10 years? Mostly yes, but rebalanced. The connectivity backbone (HFC + fiber + wireless MVNO + small-business fiber) will still be selling internet access — but to fewer households, with more wireless attach, and probably with a different subscriber-vs-ARPU mix than today. Theme parks will be larger; Universal Beijing and Epic Universe will have matured and at least one new park likely opened. Studios will still make hits and misses. Peacock will either be merged/absorbed into a larger streaming entity or shut down — sub-scale streaming services don't survive a decade.

Will the customer base be larger? Almost certainly not in residential broadband — best case flat, base case down 10-20%. Larger in wireless and BSC. Larger in theme park visitors. Net customer count probably flat-to-slightly-down.

Will profit per customer be higher? Yes in BSC, yes in wireless (as MVNO economics improve and 5G CBRS offload reduces wholesale cost), uncertain in residential broadband (ARPU mix shift up via wireless bundling, but headline broadband ARPU under pressure).

Will the moat be wider? No — narrower in residential broadband (more competition), wider in theme parks (Epic Universe entrenches), unchanged in BSC and content.

Single biggest threat: a fiber-to-the-home buildout that crosses 80% of Comcast's footprint coupled with fixed-wireless capturing the bottom-30% of households. This forces Comcast into either price competition (margin destruction) or accelerated fiber overbuild capex (cash flow destruction). Either way owner earnings step down meaningfully.

The 10-year picture is plausibly fine — a slowly-shrinking utility plus a growing parks business plus optionality on content — and plausibly grim if the broadband bear case plays out at the high end. Confidence is genuinely middling. The valuation does not require me to be right on the optimistic case; flat owner earnings at 6.5x earnings is still a great outcome. But it does require me to not be wrong on the secular case.

CONFIDENCE: medium

Position Guidance

• Recommendation: Buy
• Conviction: medium
• Target buy price: $32 (below this, margin of safety becomes meaningful even on a bearish IV revision)
• Target trim price: $55 (approaches mid-IV after stress-testing scorer range; above this, you are paid for execution rather than for cheapness)
• Position sizing: 2-4% of portfolio. Not a 'pound the table' position — secular subscriber decline and dual-class governance argue against concentration. Size up only if domestic broadband revenue stabilizes at -2% or better for two consecutive quarters.
• What kills the thesis: broadband revenue declines accelerating to -5% or worse for two consecutive quarters; or material adverse change in MVNO agreement with Verizon; or another large-cap M&A deal that resembles Sky.
• What strengthens it: wireless attach reaching 25%+ of broadband subs; BSC growth re-accelerating; Epic Universe IRR exceeding 15%; further simplification (Sky carve-out / sale).