Comcast Corporation (NASDAQ: CMCSA) — A Cash Machine Priced for Permanent Decline

Independent Equity Research · As-of date: 2026-06-12 Price: ~$24.50 · Shares out: ~3.56B · Market cap: ~$87.5B · Net debt: ~$85B · Enterprise value: ~$173B Trailing P/E: ~4.8x (one-time-inflated) · Normalized fwd P/E: ~7x · EV/Adj-EBITDA: ~4.6x · Dividend yield: ~5.4% (26% payout) · Composite valuation percentile (own 10-yr history): ~2.4th

⚡ Claude’s Take

This block is the author’s own independent opinion and general information only — not investment advice. The detailed analysis that follows takes no position and carries no price target; it discusses valuation only as embedded expectations.

Verdict: BUY / accumulate-on-weakness — a deeply-mispriced, disciplined cash machine where the market is paying you ~8–9% a year (5.4% dividend + ~5%/yr buyback) to wait out a broadband decline it is over-extrapolating, while assigning roughly zero value to a genuinely world-class theme-park duopoly. Directional fair-value zone ~$30–38 (≈7–9x normalized EPS / a conservative sum-of-the-parts), versus ~$24.50 today — ~25–55% upside to fair value before counting the cash you collect en route. This is contrarian value, not a compounder and not a falling knife.

The variant question on Comcast is not business quality — the business is, segment-by-segment, exactly what it appears: an eroding-but-still-40%-margin broadband utility, a fortress theme-park franchise, a sub-scale streamer turning the corner, and a no-moat MVNO that happens to be the best defensive weapon cable has. The variant question is price × durability, and at the ~2nd percentile of its own ten-year valuation history the market has resolved that question maximally bearishly: ~4.6x EV/EBITDA and ~7x normalized earnings price the entire enterprise as if it were levered, single-product, melting cable — when ~40% of segment EBITDA sits in structurally better or different businesses (Business Services at a 56% margin, Theme Parks in a Disney/Universal duopoly, both growing). The headline 4.8x trailing P/E is a trap — it is inflated by a one-time ~$9.4B Hulu-sale gain; the honest multiple is ~7x. But ~7x for a business throwing off ~$12–13B of real, cash-backed free cash flow behind an investment-grade, 2.3x-levered, 4.0%-coupon balance sheet, retiring ~5–6% of its shares a year, is genuinely cheap.

The reasons it is cheap are real and I am not dismissing them: broadband lost 711K subscribers in 2025 and management is choosing to cut ARPU (−3.1%) to defend units, ROIC (~8.5%) only narrowly clears WACC (~7%), the inflection is forward-dated to H2-2026/2027 and rests on an unproven “free-line-to-paid” conversion, and the Roberts family’s non-dilutable 33% super-voting control means no activist will ever force the value-unlock — you are a permanent passenger. That last point is why this is a value call, not a high-conviction one: the catalyst is self-help execution and multiple-normalization, not a corporate-action gun to management’s head. Conviction: medium. The single piece of evidence that flips me decisively bullish: two-to-three quarters of genuinely stable broadband units (organic, ex-event-noise) with ARPU re-inflecting as free lines convert — proof the convergence bet works. The single piece that flips me bearish: broadband EBITDA resuming an accelerating decline (units and ARPU both falling) with Epic/parks failing to ramp — i.e., the melt spreading from units to the whole C&P profit pool. Tag: “The market is pricing the pipes for a funeral and giving away the parks.”

1. Executive Summary

Comcast is the largest US cable-broadband operator (Xfinity, ~31.3M broadband customers across 65M homes/businesses passed) bolted onto NBCUniversal (NBC/Telemundo broadcast, Peacock streaming, Universal/Illumination studios, and the Universal theme-park empire including the May-2025-opened Epic Universe) and Sky in Europe. Following the January 2, 2026 spin-off of Versant Media (the declining cable-TV networks — USA, CNBC, MSNBC, Syfy, E!, Golf), the company reports in two groupings: Connectivity & Platforms ($80.9B revenue, $32.4B Adjusted EBITDA in FY2025) and Content & Experiences ($45.6B revenue, $6.5B Adjusted EBITDA). Consolidated FY2025: revenue $123.7B (flat), Adjusted EBITDA $37.4B, net income $20.0B, ~3.71B diluted shares.

The central tension. This is a mature, no-revenue-growth cash machine whose largest profit pool — residential broadband — is in structural subscriber decline (−711K in 2025, penetration of homes-passed falling 49.8%→47.6%) under a two-front capital-cycle assault from fiber overbuilders (AT&T, Verizon-Frontier, T-Mobile JVs; ~55% of Comcast’s footprint now overlapped) and fixed-wireless access (T-Mobile/Verizon FWA). The crown-jewel broadband EBITDA margin slipped 56.7%→55.9% and management is now voluntarily cutting broadband ARPU (−3.1% in Q1-2026) to stabilize units. The market has responded by compressing the multiple to the ~2nd percentile of its own decade — a valuation that prices in permanent, accelerating decline of the whole enterprise.

What the bear case under-weights. Roughly 40% of segment EBITDA is not residential broadband: Business Services Connectivity earns a 55.9% margin and is growing; Theme Parks earns a 31.3% margin, just opened Epic Universe, and operates in a genuine Disney/Universal duopoly with high barriers — it crossed $1B of quarterly EBITDA for the first time in Q4-2025 (+24%) and grew 33% in Q1-2026. Peacock (46M subs) reaches profitability in Q2-2026. Xfinity Mobile added a record 435K lines in Q1-2026. The company manufactures mid-single-digit EPS growth out of zero revenue growth by retiring ~5–6% of shares annually, fully funded by ~$12–13B of clean free cash flow, behind a fortress investment-grade balance sheet (net debt/EBITDA ~2.3x, 4.0% blended coupon).

Quality-of-earnings flag. The headline trailing P/E of ~4.8x is an artifact: FY2025 EPS of $5.41 was inflated by a non-recurring ~$9.4B pre-tax gain on the sale of Comcast’s Hulu stake to Disney. On clean ~$3.50 earnings (which reconciles to the ~$3.56 forward consensus) the multiple is ~7x — still cheap, but not 5x-cheap. ROIC of ~8.5% clears an estimated ~7% WACC only thinly, and the spread is compressing.

Governance. Comcast is permanently controlled by the Roberts family through a non-dilutable 33⅓% super-voting Class B stake held on ~1% of the economics. The mitigants are real (82% independent board, strong Lead Independent Director, a genuinely shareholder-aligned ROIC/EPS/TSR incentive plan, no history of minority expropriation) but the structural fact remains: no activist, hostile bid, or proxy contest can ever force a re-rating. The catalyst must come from execution.

No recommendation or price target appears below this section (the only opinion is the Claude’s Take block above). The body analyzes valuation strictly as embedded expectations and scenarios.

2. Business Overview

Comcast makes money four ways: selling connectivity (broadband, wireless, voice) to ~50.8M residential and business customer relationships; selling video/entertainment subscriptions and advertising (Sky, Peacock, NBC, Telemundo); producing and distributing film and television (Universal, Illumination, Focus, Sky Studios); and operating destination theme parks. Following the 2025 segment reorganization and the January 2026 Versant spin-off, the reporting structure is:

(FACT — FY2025 10-K, filed 2026-02-03, MD&A segment tables; Adjusted EBITDA is Comcast’s primary segment profit metric.)

Revenue model — recurring vs. cyclical. The clear majority of revenue (INTERPRETATION: ~60%+) is subscription/recurring: broadband, business connectivity, Sky, Xfinity Mobile service, Peacock, and theme-park season passes. This is the ballast that makes the cash flows so predictable. Layered on top are two important cyclical/lumpy streams: (1) Media advertising, which carries a structural even-year bias from the Olympics + US political-election cycle — the Paris 2024 Olympics flattered FY2024 Media revenue, and FY2025 gave it back; Q1-2026 then benefited from a “Legendary February” (Milan-Cortina Winter Olympics + Super Bowl LX + NBA All-Star) worth ~$2.2B of incremental revenue; and (2) Studios, which is hit-driven and binary (the FY2025 slate carried Super Mario, which grossed >$750M globally; a quiet slate year looks very different). Theme Parks are discretionary and macro/travel-sensitive but largely recurring within a year.

Customer KPIs (FY2025, FACT — 10-K): total Connectivity & Platforms customer relationships 50.8M (−967K YoY); domestic broadband customers 31.3M (−711K); domestic wireless lines 9.3M (+1.5M); domestic video customers 11.3M (−1.3M); domestic homes/businesses passed 65.0M (+1.3M); broadband penetration of passings 47.6% (down from 49.8%). The single most important operating fact in the entire memo is in that list: the broadband base is shrinking and penetration of an expanding network is falling — the textbook signature of an oversupplied market.

Verdict. A predictable, cash-generative, subscription-heavy conglomerate whose center of gravity is connectivity (65% of revenue, ~83% of segment EBITDA) with two genuinely differentiated experiences assets (Parks, Business Services) and a structurally-hard content/streaming wing. The 2025 reorganization and Versant spin were a deliberate effort to re-present the company as “six growth drivers plus a melting legacy” — accurate as far as it goes, but the growth drivers do not yet outweigh the decline in the core.

3. Industry Dynamics

Comcast straddles four industries of sharply different structural quality. A blended verdict requires sizing each by profit pool.

US residential broadband — the dominant profit pool, structurally bad and oversupplied. This is where ~$27B of the ~$32B Connectivity & Platforms EBITDA sits, and it is at the wrong end of the Marathon capital cycle. The setup: cable earned high, stable, near-monopoly returns on already-sunk hybrid-fiber-coax plant for two decades. Per the iron logic of the capital cycle, high returns attract capital — and capital is now flooding in, not from new cable entrants but from two adjacent industries with their own balance sheets and subsidies:

• Fiber overbuild (the durable, ceiling-less threat). AT&T is pushing from ~30M toward 60M+ fiber passings by 2030; Verizon closed its acquisition of Frontier in January 2026 (>30M combined fiber homes); T-Mobile is building via off-balance-sheet JVs (Lumos, Metronet). Comcast’s footprint is ~55% fiber-overlapped today, heading toward ~60% (FACT — Q1-2026 call). Once built, fiber is a strictly superior product (symmetrical multi-gig, lower latency) at comparable price, and unlike FWA it has no capacity ceiling. The $42.45B federal BEAD program subsidizes still more rural overbuild — of competitors, not incumbents.
• Fixed wireless access (the capacity-capped but disruptive threat). T-Mobile and Verizon converted spare 5G capacity into >13M FWA broadband subscribers, capturing ~32% of US broadband gross adds in Q4-2025 (cable fell to ~41%, its lowest ever; fiber ~26%). New Street now estimates the carriers can support ~32M FWA subs (roughly double its mid-2024 estimate, lifted by AT&T’s ~$23B EchoStar spectrum purchase). But FWA net adds are decelerating (Verizon’s slowed from 308K to 214K YoY into Q1-2026, and it is explicitly shifting mix “more toward fiber”), and FWA is a best-effort product that cannot durably serve the high-usage core. (Leichtman Research; New Street via Fierce Network.)

The category went from +5.2M net adds in 2020 to roughly −1.3M in 2024 — a saturated, ex-growth end market with capital pouring in. This is a structurally bad industry, and the Greenwald warning applies in real time: “market growth and new capital are the enemy of economies-of-scale advantages.” The capital-cycle counter-argument (the bull’s hope) is that cable is exiting its own capex cycle (DOCSIS 4.0 build completing) while the telcos are mid-boom, and the lowest-cost-per-bit survivor harvests cash once entrant capital exhausts and FWA’s ceiling binds. That is a “when,” not an “if,” and it is genuinely the crux.

Pay-TV — structurally terrible, in secular collapse (Comcast lost 1.3M video subs in 2025). Correctly de-emphasized; the worst of it (cable networks) was just spun into Versant.

Streaming — scale-driven, capital-intensive, winner-take-most. Peacock (46M subs) is sub-scale against Netflix (~300M+) and Disney+; structurally disadvantaged on content-budget-per-sub. A hard industry in which Comcast is a credible #4-5, not a leader.

Theme parks — structurally excellent. A near-duopoly of Universal and Disney protected by the highest entry barriers in entertainment: thousands of acres of entitled land near tourism hubs, $1B+ per-attraction capital intensity, decades of operating know-how, and irreplaceable IP. Pricing power, ~30%+ margins, secular attendance growth tempered only by cyclicality.

Industry verdict: structurally mixed-leaning-poor on a revenue-weighted basis. The dominant profit pool (broadband) is in a bad, oversupplied, ex-growth industry; this is offset by two structurally good islands (Business Services, Theme Parks) and one structurally hard bet (streaming). The crux of the entire thesis: the market’s ~4.6x EV/EBITDA prices the whole company as the bad-broadband industry, when ~40% of EBITDA is in better or different industries.

4. Competitive Position

The broadband moat is real but eroding — name it precisely. Comcast’s advantage on an already-passed home is a local economies-of-scale cost advantage on sunk-cost plant, in Greenwald’s taxonomy. The HFC network passing 65M premises is built and largely depreciated; the marginal cost to connect and serve one more home is near-zero. An entrant must sink fresh capital — ~$1,000–1,500/home for fiber, or scarce spectrum capacity for FWA — to contest revenue Comcast already captures at almost no incremental cost. That cost asymmetry is the entire moat. It is demonstrably real: you do not earn a 40% segment margin (or Business Services’ 56%) on a commodity without a structural cost advantage.

But the moat has a fatal feature in the current environment: its customer-captivity leg is weak. Broadband is a near-commodity with low switching costs; customers historically stayed not because they were locked in but because there was no cheaper-to-serve competitor at the home. Fiber overbuild and FWA dismantle exactly that premise. The financial evidence of erosion is unambiguous: broadband customers −711K in 2025 (accelerating from −411K in 2024), penetration of passings 49.8%→47.6%, and — most tellingly — broadband ARPU turned negative (−3.1% in Q1-2026), meaning the pricing power that was historically part of the moat is now gone. This is erosion, not collapse: the moat defends a high share floor and harvests cash, but it no longer delivers unit growth or pricing power.

The theme-park moat is genuinely strong and durable — intangible IP (Harry Potter, Nintendo’s Super Nintendo World, Jurassic, Illumination’s Minions) + irreplaceable location + economies of scale, in a duopoly with Disney. Financialized in a 31.3% margin and Epic Universe’s ramp (Q1-2026 parks revenue +24%, EBITDA +33%). This is the one segment with a wide, durable, financialized moat — and the one the cable multiple ignores. Disney itself cited “Epic-related headwinds” to domestic attendance in its most recent quarter — Universal is taking share in a structurally good industry.

Studios/Media have weak-to-no durable moat — hit-driven film, a melting linear annuity, and a sub-scale streamer. The one genuine asset is the content-rights portfolio (Olympics through 2036, NBA, NFL/SNF, Premier League), which is durable but expensive and currently margin-dilutive.

Xfinity Mobile has no moat — it is an MVNO reselling Verizon’s network (with a T-Mobile MVNO launching for business). Zero facilities-based advantage. Yet it is strategically the most important offensive/defensive weapon Comcast has: it deepens the broadband relationship, lowers churn, and raises lifetime value, with ~90% of traffic offloaded to Comcast’s own WiFi. It added a record 435K lines in Q1-2026 at 16% penetration of the broadband base, with convergence ARPA ~$85 versus telecom competitors’ ~$170 — a long monetization runway if the free-line cohorts convert to paid.

Head-to-head. Versus Charter (closest cable peer): Comcast is far less levered (2.3x vs. Charter’s ~4.2x — a critical distinction, Comcast is not a melting-ice-cube balance sheet) and more diversified (parks/studios Charter lacks); Charter has higher mobile penetration. Versus AT&T / Verizon-Frontier (fiber): losing share where fiber arrives, but to a high floor and slowly, defended by DOCSIS 4.0. Versus T-Mobile/Verizon (FWA + wireless): losing the price-sensitive broadband tail to FWA while countering hard on the mobile side. Versus Disney/Netflix: a co-equal parks duopolist (offense) and a distant streaming also-ran (defense).

DOCSIS 4.0 is the technical linchpin: Comcast has deployed full-duplex 4.0 amplifiers across its footprint, offering symmetrical multi-gig to “millions” of homes at a small fraction of a greenfield fiber build’s cost-per-home. This closes most of the product gap with fiber while preserving the cost gap — and if a 2Gbps DOCSIS connection is indistinguishable from 2Gbps fiber, the overbuilder’s sunk capital earns a far lower return (the “ROI fatigue” the bull thesis requires). The honest read: DOCSIS 4.0 achieves speed/symmetry parity at a structural cost advantage — enough to defend a high share floor and harvest cash, not enough to restore broadband unit growth or pricing power.

Competitive verdict: differentiated by segment, and the bear case over-generalizes. This is not a single-product melting ice cube — it is a partially-eroding utility moat (real, financialized, dissolving home-by-home to a high floor) wrapped around a fortress parks moat, priced as if it were all the former.

5. Growth History and Forward Opportunities

History (organic — there has been no material acquisition in the window). Consolidated revenue has been dead-flat at ~$121–124B for four years (a ~0.5% CAGR FY2022–2025). Beneath that flat line, a bifurcation: management’s “six growth drivers” (broadband ARPU until 2025, wireless, Business Services, Theme Parks, Studios in good years, Peacock) grew from ~50% to >60% of revenue, while the legacy base (video, linear Media) declined. The genuine growth came from:

• Xfinity Mobile: lines from 7.8M to 9.3M in 2025 (+19%), service revenue +15–20%.
• Theme Parks: EBITDA crossed $1.0B in a single quarter for the first time in Q4-2025 (+24% YoY); Japan posted its second-best EBITDA year ever.
• Peacock: revenue grew to $5.4B (+10%) with subs ~36M → 44M (YE2025) → 46M (Q1-2026); losses narrowing from ~$2.5B (2023) to ~$1.8B (2024) to ~$1.1B (2025).
• Business Services: steady ~6% revenue growth (EBITDA growth decelerating to 3–4% as SMB is pressured by FWA).

Where it shrank: video (−1.3M subs), linear Media (now spun into Versant), and — newly and by management’s own choice — broadband ARPU, which flipped from +1.1% (Q4-2025) to −3.1% (Q1-2026) as Comcast launched simplified all-in pricing, a five-year price guarantee, and a 12-month free wireless line.

The broadband-subscriber trend — the single chart that drives the stock (domestic broadband net adds/losses, thousands):

(FACT — quarterly earnings releases.) The trend deteriorated through 2024 into early 2025, then began to flatten — the Q1-2026 −65K is the best print since the losses began, though management itself flagged that >50% of the YoY improvement was the non-repeatable “Legendary February” event, so the durable organic run-rate is a more modest ~−110 to −115K. The bull needs this series to keep flattening toward zero; the bear expects it to re-deteriorate as fiber overlap climbs from ~55% toward ~60%.

The result is an investment-trough year, not a growth year, on reported numbers: consolidated Adjusted EBITDA declined ~9–10% in both Q4-2025 and Q1-2026, as the growth businesses are not yet large enough to offset the broadband-ARPU reset plus peak NBA-rights amortization.

Forward opportunities, ranked by quality:

1. Theme Parks (cleanest runway). Epic Universe (opened May 2025) will be “fully ramped” only by end-2026 — embedded upside remains — and is lifting the entire Orlando resort (hotel ADR +20%, +2,000 rooms, higher per-cap spend, “week-long destination”). The pipeline is multi-year: Universal Kids Resort Frisco TX (summer 2026), a new Hollywood coaster (2026), Universal UK Bedford (groundbreaking 2026, opening ~2031), Horror Unleashed Las Vegas. Plausibly a mid-to-high-single-digit-plus EBITDA grower for years (INTERPRETATION), tempered by continued capex and soft international parks (Osaka/Beijing).
2. Xfinity Mobile / convergence (most-emphasized, partly unproven). Record line adds, ~half from 12-month free lines whose cohorts begin converting to paid in H2-2026 — management claims a “vast majority” convert (FACT but unquantified; this is the single biggest swing factor). Premium plans (~30% of connects) and the new Mobile Plus bundle (lifetime device protection) extend reach upmarket.
3. Broadband stabilization (contested). Q1-2026’s −65K loss improved +117K YoY (first YoY improvement since Q4-2020) — but management disclosed that >50% came from the non-repeatable “Legendary February” marketing surge, so the organic improvement is ~50K. Real but partly borrowed; against rising fiber overlap and now Starlink.
4. Peacock profitability. Confirmed profitable in Q2-2026 (Evercore, June 2026) — a first. Bundling (Apple, Walmart, Amazon, Roku) is “the next wave.”

Growth verdict: a genuine high-quality growth core (Parks, wireless line momentum, Peacock-to-profit) is temporarily masked by a self-inflicted broadband-ARPU reset and peak NBA dilution. The mix is improving; the consolidated line is, for now, low-quality. The bull case requires the H2-2026 free-line monetization to deliver at scale and at acceptable ARPU — currently asserted, not proven.

6. Financial Quality

The five-year financial arc, in one table:

*FY2022 depressed by the ~$8.6B Sky impairment. **FY2025 inflated by the ~$9.4B Hulu-sale gain (see Quality-of-earnings below). The two numbers that tell the whole story: revenue was dead-flat for four years while the diluted share count fell ~20%. That is the equity engine in one line — manufacturing per-share growth by shrinking the count against a stagnant top line. Note also the GAAP-EPS volatility (Sky impairment 2022, Hulu gain 2025) that makes any single-year P/E unreliable and is precisely why the market’s headline “4.8x” is so misleading.

Revenue and margins. Flat revenue (~$123.7B FY2025) with eroding margins: GAAP operating margin fell from 19.2% (2023) to 16.7% (2025); the crown-jewel Residential C&P Adjusted EBITDA margin slipped 56.7%→55.9%. Consolidated Adjusted EBITDA margin ~30.2%. The high-margin broadband core is being defended with rising capex even as its margin slips — Connectivity & Platforms capex rose 5.3% to $8.7B. Scale is not producing incremental margin; it is being spent to hold a shrinking base.

Free cash flow. Here lies the most important quality nuance. Reported FY2025 operating cash flow was a record $33.6B, but it is inflated by Hulu-sale proceeds/tax timing and working-capital swings — it is not a clean run-rate. Comcast’s underlying, sustainable free cash flow (CFO less ~$11.75B capex and ~$2.6B intangibles) is ~$12–13B, consistent with FY2022–2024. Capex intensity (~9.5% of revenue) was elevated FY2023–2025 by the Epic Universe build layered on ~$8.7B of network capex; Epic now rolls off — but, crucially, management is redeploying that relief into a record 2026 broadband-investment year rather than harvesting it, so near-term FCF expansion is muted.

Balance sheet — a genuine strength. Total debt ~$94.6B (Q1-2026), cash ~$9.5B, net debt ~$85B, net-debt/EBITDA ~2.3x — moderate, stable, investment-grade leverage at a 4.0% weighted-average coupon with laddered maturities and hedged FX. Interest expense (~$4.0–4.1B/yr) is covered ~9x by Adjusted EBITDA. There is no refinancing cliff or liquidity risk. This is what separates Comcast from the levered single-product cable names: it can comfortably fund the network defense, a growing dividend, and large buybacks simultaneously.

Quality-of-earnings — the headline P/E is a trap. FY2025 diluted EPS of $5.41 (+30% YoY) was inflated by a ~$9.4B pre-tax (~$7B after-tax) non-recurring gain on the sale of Comcast’s Hulu interest to Disney (booked Q2-2025; quarterly net income spiked to $11.1B). Strip it and underlying EPS is ~$3.50 — which reconciles almost exactly to the ~$3.56 forward consensus. The “trailing P/E of ~4.8x” is therefore an illusion; on clean earnings the multiple is ~7x. Other items to normalize: the FY2022 Sky impairment (~$8.6B, depressing that year’s base and flattering subsequent growth); Olympics/election cyclicality in Media (compare two-year stacks); a FY2024 tax benefit (effective rate 15.0%, normalizing to 23.7% in 2025); and the cosmetic January 2026 share-ratio adjustment associated with the Versant distribution. Net income is cash-backed in a normal year (clean CFO ~$28B vs. clean NI ~$13B, the gap being ~$13B D&A) — no accrual red flag.

Returns. Reported ROE ~22% is doubly flattered — by the Hulu gain (numerator) and by years of buybacks holding equity flat (denominator); normalized ROE is ~14–15%. The honest measure for a levered, buyback-heavy structure is ROIC ~8.5% (NOPAT ~$15.7B on ~$185B invested capital), which clears an estimated WACC of ~7–7.5% by only ~100–150bps — and the spread is compressing as broadband margins erode and capex stays elevated. P/B is ~1.0x, so buybacks are being done near book value — value-conscious, not the deep-discount repurchase the headline P/E might suggest.

Financial-quality verdict: do economics improve with scale? No — economics are flat-to-deteriorating despite massive scale. This is a mature cash cow with a narrowing return spread, not a compounder. What keeps it investable is the financial machine — ~$12–13B of clean, cash-backed FCF behind a fortress balance sheet, funding a buyback that retires ~5–6% of shares a year and manufactures EPS growth out of zero revenue growth. The single most important quality flag is that the headline 4.8x P/E is inflated by the Hulu one-timer; on clean ~$3.50 EPS the multiple is ~7x — cheap, but not 5x-cheap.

7. Capital Allocation

M&A — a barbell. The signature error of the Roberts era is the ~$39B Sky acquisition (2018), outbid against Fox at a media-cycle top, written down ~$8.6B in 2022 — a canonical Marathon “high returns invite capital at the top” misstep. But the recent record is disciplined subtraction-as-strategy: the Hulu monetization (~$9.4B, cleanly settling a long-running put/call), the tax-free Versant spin (shedding the worst-melting cable networks while retaining NBC/Telemundo/Peacock/Bravo), and — tellingly — Comcast walking away from Warner Bros. Discovery in 2025 “at these values” when it went all-cash. The bias has shifted decisively from empire-building to pruning. Verdict: mixed, improving — the Sky scar is permanent, but the current playbook is rational.

Capital returns — the genuine strength. Dividends per share grew $1.08 (2022) → $1.32 (2025), a 19-consecutive-year growth streak (though the cash rate was held flat for 2026 — a notably softer posture, with the Versant in-kind distribution counted as the “growth”). Buybacks ran $13.3B (2022) → $11.3B → $9.1B → ~$7B (2025) as Epic capex and dividends competed for cash, retiring ~16% of shares in three years (~5–6%/yr) near ~1.0x book. A $15B authorization was reset in January 2025. Total shareholder yield is ~8–9% — a high, FCF-funded floor of return manufactured against a no-growth top line.

Capex allocation splits across network (DOCSIS 4.0 — defensive/retention capex, not growth), Epic Universe (~$7B, the largest discretionary bet, early signs accretive but full-cycle ROI unproven, with UK/Frisco parks to come), and content. The honest caveat: the “capex rolls off post-Epic, FCF expands” thesis is only half true — management is redeploying the relief into broadband, not harvesting it.

Governance — permanent founder control. The Roberts family holds 9.44M Class B shares (15 votes each) equal to a non-dilutable 33⅓% of combined voting power on a ~1% economic stake — the classic dual-class wedge. The mitigants are real: 82% independent board, a strong Lead Independent Director (Edward Breen), no insider pledging/hedging, a clean succession (Mike Cavanagh elevated to co-CEO January 2026), and — critically — no documented history of self-dealing or minority expropriation. But the structural fact is decisive: no activist, hostile bid, or proxy contest can ever dislodge the family or force a value-unlock. The same control structure was replicated in Versant. Minority holders are permanent passengers; alignment runs through incentives and the buyback, not through external accountability.

Incentives — better than typical for a controlled company. The long-term PSU plan (100% of 2025 equity grants) is weighted equally on average absolute ROIC and relative Adjusted-EPS-growth CAGR vs. the S&P 100, with a relative-TSR ±25% modifier capped so no positive modifier applies if absolute TSR is negative. The annual bonus is tied to revenue, Adjusted EBITDA, and free cash flow. This pays management for per-share value creation and capital efficiency, not empire-building — directly addressing the Greenwald/Marathon concern. Brian Roberts’ total comp is stable at ~$35M (~67% performance-based); say-on-pay support was 90%.

Insider behavior — neutral-to-mildly-negative. There are zero open-market purchases (code P) in the entire five-year Form 4 corpus — not a single conviction buy by any officer or director, even with the stock near a decade-low. Activity is 100% routine (option exercises, tax withholding, grants, gifts). Insiders own 1.13% of Class A. There is no insider-buying tailwind.

Capital-allocation verdict: competent, disciplined, and improving — but permanently entrenched and scarred by one large misstep. The current playbook (prune declining media, defend the network, return ~8–9%/yr at a low multiple, hold leverage modest, pay on ROIC/per-share metrics) is rational and shareholder-conscious. But honesty requires stating that this is a founder-controlled entity where the market-for-corporate-control discipline does not exist, and that the same management overpaid catastrophically for Sky within the window.

8. Changes and Headwinds — Last Two Years

Headwinds: (1) broadband sub erosion against ~55%-and-rising fiber overlap, FWA, and now Starlink — management explicitly plans assuming the environment does not ease; (2) self-inflicted broadband-ARPU pressure (one more quarter guided); (3) NBA amortization drag through 2028; (4) Peacock durable profitability still to be proven across NBA-heavy quarters; (5) theme-park cyclicality and soft international parks (China-related inbound decline); (6) linear/Media secular decline; (7) a frozen dividend cash rate and a 2026 FCF step-down (2025 had a ~$2B one-time cash-tax benefit and Versant FCF now removed); (8) a leverage tick-up post-Versant that bounds buyback capacity.

Verdict: the 2-year changes net-strengthen the portfolio mix and capital discipline but do not yet strengthen reported earnings — the inflection is deliberately forward-dated to H2-2026/2027 and rests on unproven execution.

9. Risk Analysis

The risk profile is dominated by slow structural erosion of the broadband core, not by any acute solvency or single-event risk. The balance sheet and cash generation make a permanent capital impairment highly unlikely; the real risk is a value trap — the multiple stays washed-out because broadband keeps melting faster than parks/wireless can offset, and no catalyst ever forces a re-rating.

10. Valuation Discussion (Embedded Expectations)

No price target or recommendation here (the only opinion is the Claude’s Take block at the top). This section frames what the current price implies.

The multiples. At ~$24.50, Comcast trades at ~4.6x EV/Adjusted-EBITDA, ~7x normalized forward earnings (the ~4.8x trailing figure is Hulu-inflated), ~0.70x sales, a ~5.4% dividend yield at a 26% payout, and — most strikingly — the ~2.4th percentile of its own ten-year valuation history (P/E 5th, P/B 1.6th, P/S 0.3th percentile; own-history percentiles, compared only against the stock’s own past). On clean FCF of ~$12–13B against an ~$87.5B market cap, the free-cash-flow yield is ~14–15%.

What the price embeds. A ~14–15% FCF yield on a business with an investment-grade balance sheet is an extraordinary implied required return. Decomposed: for the equity to be “fairly” valued at, say, a 9% FCF yield (reasonable for a low-growth, mildly-declining cash generator), FCF would need to fall to ~$8B — i.e., the market is pricing a ~35%+ permanent decline in free cash flow. Equivalently, holding FCF flat, the market is demanding a ~14% forward return, which only makes sense if it expects meaningful terminal value-destruction. The embedded expectation is that the broadband melt spreads to the whole enterprise and compounds — that parks, Business Services, and wireless do not, in aggregate, offset the decline of the core.

Sum-of-the-parts — the variant-perception engine. Because the segments belong to different industries, a blended EV/EBITDA understates the whole if the pieces deserve different multiples:

(INTERPRETATION — illustrative, not a target; multiples chosen conservatively. Parks alone, at ~$3.1B EBITDA and a Disney-comparable mid-teens multiple, would be worth ~$30–45B — meaningful against an ~$87.5B equity value.) The point is not precision; it is that even applying depressed cable multiples to the connectivity core and modest multiples to the rest, the parts sum to materially more than the whole — the market is assigning little-to-no value to the parks duopoly and the 56%-margin Business Services franchise.

The bear’s valuation counter. Apply 4.0x to a declining C&P EBITDA ($130B), 9x to parks ($28B), 5x to media/studios ($17B) → $175B EV − $85B net debt → ~$90B → ~$25/share. In other words, the current price is defensible only if you believe C&P EBITDA enters a sustained decline (not just flat) and the parts deserve trough multiples. The valuation debate reduces to one question: is Connectivity & Platforms EBITDA flat-to-slowly-declining (the price is too low) or in sustained acceleration toward decline (the price is fair)?

Scenario sketch (illustrative):

• Bear: C&P EBITDA −3–4%/yr indefinitely, parks de-rate, no convergence payoff → FCF drifts toward ~$9–10B, multiple stays ~4x → ~$18–22/share.
• Base: C&P EBITDA roughly flat (ARPU recovery as free lines convert offsets unit declines), parks ramp, Peacock profitable, buyback continues → ~$12–13B FCF, modest re-rate to ~6–7x EBITDA / ~9x earnings → ~$30–34/share.
• Bull: broadband units stabilize and ARPU re-inflects, parks compound, modest multiple normalization toward the historical mean → ~$36–42/share.

You are paid ~8–9%/yr (dividend + buyback) to wait for that resolution.

11. Variant Perception

Consensus is genuinely ambivalent — analyst ratings cluster around hold (mean ~3.7/5; roughly 13 buy, 14 hold, 2 sell), short interest is a benign ~2.2% of float (this is not a crowded short — it is a neglected, washed-out value name), and the recent tape is quiet (Rosenblatt cut its target to $24 in June 2026). The consensus view: a well-run but structurally challenged cable/media conglomerate whose broadband core is in slow secular decline, cheap for a reason, with no obvious catalyst — a value trap to be avoided or held, not bought.

The bull case (the variant): the market is over-extrapolating broadband melt and pricing the entire company as deteriorating cable, when ~40% of EBITDA sits in structurally better or different businesses (a parks duopoly, a 56%-margin Business Services franchise) that are growing. On clean ~$3.50 earnings the stock is ~7x with a ~14% FCF yield and an 8–9% shareholder yield; a sum-of-the-parts that values parks even modestly implies ~$30–38. The broadband decline is real but bounded (DOCSIS 4.0 defends a high floor; FWA is capacity-capped and decelerating; the cable capex cycle is ending while telcos’ is mid-boom), and the inflection is forward-dated but plausible (Epic ramp, Peacock profitability, free-line monetization). You are paid handsomely to wait.

The bear case: broadband is in secular, accelerating decline that no amount of DOCSIS or convergence will arrest; the ARPU cut is an admission the moat is gone; ROIC barely exceeds WACC and is compressing; parks are cyclical and Epic’s ROI is unproven; the Roberts family entrenchment means no activist will ever force the value-unlock, so the discount is permanent; and management just froze the dividend — a tell. Cheap stays cheap.

The 3–5 assumptions that decide it:

1. Does broadband C&P EBITDA stabilize or keep declining? (The whole valuation hinges here.)
2. Do the free wireless lines convert to paid at acceptable ARPU? (The convergence payoff.)
3. Does Epic Universe / parks deliver the EBITDA ramp and re-rating the SOTP implies?
4. Is FWA’s damage largely done (capacity-capped, decelerating) or just beginning (ceiling rising)?
5. Will the founder-controlled structure ever permit — or the multiple ever reflect — a re-rating absent an external catalyst?

Falsification: the bull is falsified by two-plus quarters of broadband units and ARPU both declining with parks failing to ramp (the melt spreading to the whole C&P pool). The bear is falsified by genuine organic broadband-unit stabilization with ARPU re-inflecting as free lines convert — proof the convergence moat-defense works.

12. Fact vs. Interpretation Table

13. Open Questions

1. Clean run-rate FCF post-Versant and post-Epic — Versant removes some EBITDA while Epic capex is redeployed into broadband; net direction needs FY2026 guidance.
2. Versant carve-out magnitude — how much RemainCo FCF/EBITDA the spin actually removed (FY2025 figures still include Versant).
3. Sustainability of the Q1-2026 broadband improvement — seasonal + event-driven + residential-only reclassification; watch Q2–Q3 2026 for the organic run-rate.
4. Free-line-to-paid conversion economics — “vast majority convert” is unquantified; the single biggest swing factor, and the ARPU at which they convert matters as much as the rate.
5. Standalone Xfinity Mobile profitability — line growth is strong but MVNO EBITDA contribution is not separately disclosed.
6. Will the dividend resume cash-rate growth in 2027, or is the 2026 freeze a structural signal about FCF pressure?

14. What Must Be True

For the bull (cheap, mispriced value):

• Connectivity & Platforms EBITDA must stabilize (roughly flat) — ARPU recovery as free lines convert must offset continued unit attrition. Falsification: two-plus consecutive quarters of C&P EBITDA declining at an accelerating rate, with both units and ARPU falling.
• Theme Parks must deliver the Epic ramp and a multi-year EBITDA growth path. Falsification: parks EBITDA flat-to-down through 2026–2027 ex-cyclical, signaling Epic disappointed.
• The buyback (~5–6%/yr) and ~5.4% dividend must continue, funded by ~$12–13B FCF. Falsification: a dividend cut or a buyback pause for reasons other than opportunistic M&A.

For the bear (permanent value trap / melting core):

• Broadband must enter sustained EBITDA decline (units and price), with DOCSIS 4.0 and convergence failing to defend the floor. Falsification: organic broadband-unit stabilization with ARPU re-inflecting.
• FWA/fiber must keep taking share faster than parks/wireless can offset, dragging consolidated EBITDA structurally lower. Falsification: consolidated Adjusted EBITDA returning to growth in H2-2026/2027 as guided.
• The founder-controlled discount must prove permanent (no re-rating absent an unforceable catalyst). Falsification: a sustained multiple normalization toward the historical mean on operational stabilization.

15. Source Appendix

(Full source list with URLs in the Source Appendix below.) Primary sources: Comcast FY2025 Form 10-K (filed 2026-02-03); Q1-2026 Form 10-Q (filed 2026-04-23); DEF 14A proxy (filed 2026-04-24); Form 4 corpus (2021–2026); Q4-2025, Q1-2026, and December-2025 Versant Investor Day transcripts; Evercore TMT conference (June 2, 2026); EDGAR XBRL company-concept API. Industry/competitive: Leichtman Research, New Street (via Fierce Network), Light Reading. Quantitative cross-checks: public market-data sources, reconciled to filings.

The body of this article carries no investment recommendation and no price target; the only opinion is the clearly-labeled Claude’s Take block at the top, which is the author’s own independent view and general information only — not investment advice.

APPENDIX A — Standard Diligence Questionnaire

As-of 2026-06-12. Supplemental to the article above. Fact / Interpretation / Assumption labels applied where material.

General

What thoughtful questions have other investors asked about this company? The dominant question is whether broadband subscriber losses are a cyclical (competition-driven, stabilizing) or secular/terminal phenomenon — i.e., is cable’s local-scale moat eroding to a high floor or collapsing? Secondary debates: (1) whether the sum-of-the-parts (parks + business services worth far more than the cable multiple implies) will ever be reflected absent a breakup the Roberts family will never permit; (2) whether the deliberate broadband-ARPU cut (−3.1%) is smart defense or capitulation on pricing power; (3) whether Epic Universe’s ROI justifies the ~$7B build; (4) whether Peacock can be durably profitable, not just one quarter; and (5) whether the dividend freeze for 2026 signals deeper FCF stress. Activist interest has been floated (Dec 2025) but is structurally toothless given dual-class control.

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? INTERPRETATION: a self-inflicted trough on reported numbers — consolidated Adjusted EBITDA declined ~9–10% in Q4-2025 and Q1-2026 due to the voluntary broadband-ARPU reset plus peak NBA-rights amortization, both of which are designed to invert in H2-2026/2027. Underlying cash generation (~$12–13B FCF) is steadier than reported EBITDA suggests. FY2025 GAAP EPS was a one-time high (Hulu gain).

Driven by external environment or internal actions? Both: external (fiber/FWA competition pressuring broadband units) and internal/deliberate (ARPU cut, NBA deal, Epic pre-opening costs — management chose these investments).

How stable are revenues? Very stable at the top line (~$121–124B for four years, ~60%+ subscription/recurring) but stagnant. Media advertising carries even-year Olympics/election cyclicality; parks are discretionary; Studios is hit-driven.

Outlook for products/services / market size? Broadband: large (~$100B+ US) but saturated and contested — flat-to-shrinking units, defended by ARPU and convergence. Wireless: growing share for cable via MVNO. Theme parks: growing, duopoly, global (US/Japan/Europe pipeline). Streaming: large and growing but winner-take-most and capital-intensive. Net: a no-growth core wrapped around several genuine growth pockets.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? Broadband: dramatically more competitive (fiber overbuild ~55% of footprint and rising, FWA, Starlink). Theme parks: stable duopoly. Streaming: intensely competitive.

How profitable is the business (ROIC, ROE)? ROIC ~8.5% (clears ~7% WACC thinly, compressing). Reported ROE ~22% but flattered by the Hulu gain and buyback-shrunken equity; normalized ~14–15%. Use ROIC. Segment margins are strong where the moat is real (Business Services 56%, Parks 31%, Residential C&P 38%).

How profitable is the industry / barriers to entry? Broadband is a regional duopoly/oligopoly with high capital barriers but those barriers are being overcome by well-capitalized telcos. Theme parks have the highest entry barriers in entertainment. Streaming has low entry but high scale barriers.

Can the business be easily understood? Yes — connectivity utility + parks + content. Reasonably transparent segment disclosure.

Undermined by foreign low-cost labor? No — physical-network and US-domestic-experience businesses.

Do brands matter? Selectively: Xfinity is a utility brand (low loyalty); Universal/Harry Potter/Nintendo IP is a genuine, durable brand/intangible moat in parks; NBC/Peacock brands matter modestly in a crowded field.

Nature of competition / switching costs? Broadband: low switching costs (the moat’s weak leg) — historically retained by lack of a cheaper-to-serve rival, a premise fiber/FWA dismantle. Convergence (broadband+mobile bundle) is the deliberate effort to manufacture switching costs.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet? Yes — the theme-park real estate/IP and the sunk, largely-depreciated HFC/fiber network plant carry economic value well above book; the parks franchise in particular is under-represented (a SOTP argument).

Off-balance-sheet liabilities? Operating/finance leases and content/sports-rights commitments (notably the ~$2.5B/yr × 11-yr NBA deal); standard for the sector, disclosed, not alarming.

How conservative is the accounting? Reasonable. The main quality flags are normalization items (Hulu gain, Sky impairment, Olympics cyclicality, NBA straight-line amortization), not aggressive accruals. Net income is cash-backed in a normal year.

How CapEx-hungry? Moderately-to-heavily — ~9.5% of revenue (~$11.75B), split between defensive network capex (DOCSIS 4.0) and discretionary park builds (Epic, UK, Frisco). Network capex is retention, not growth.

Capital Allocation & Management

How much FCF / how used / philosophy? ~$12–13B clean FCF; used for a growing dividend (19-yr streak, ~$4.9B), buybacks (~$7B, ~5–6%/yr share reduction at ~1.0x book), and selective large growth bets (Epic). Philosophy: mature cash-cow — fund the network, return ~8–9%/yr, hold leverage ~2.3x.

Significant acquisitions recently? No acquisitions — the opposite: divestiture (Versant spin Jan 2026) and monetization (Hulu sale ~$9.4B). Walked from WBD when it went all-cash. The bias is pruning, not buying. The historical scar is Sky (~$39B, 2018 → ~$8.6B impairment).

Buying back shares? Yes, aggressively and accretively (~16% of shares in 3 years).

Issuing shares to insiders? Modest — 100% of 2025 equity grants are performance PSUs (ROIC + relative EPS + TSR-modified); not egregious dilution.

Compensation / motivations of management? Roberts comp stable ~$35M, ~67% performance-based; say-on-pay 90%. Incentives are genuinely per-share-aligned (ROIC/EPS/TSR, not empire-building) — better than typical for a controlled company. Motivation: the Roberts family’s permanent 33% voting control means a long-term, franchise-preservation orientation, but also entrenchment with no external accountability.

Valuation & Market Data

ADR / MLP / K-1? No — a US C-corp common stock (NASDAQ: CMCSA), Class A. Class B is family-held, not public. Standard 1099 dividends. Note: the Jan-2026 Versant distribution (1 VSNT per 25 CMCSA) was a taxable/return-of-basis spin event for holders.

Dividend policy? ~$1.32/share, ~5.4% yield, 26% payout, 19 consecutive years of growth — though the cash rate was held flat for 2026 (a softer posture). Well-covered, low payout, ample room.

How profitable? Highly cash-profitable (~30% Adj-EBITDA margin, ~$12–13B FCF) but only thinly economically profitable on invested capital (ROIC ~8.5% vs. ~7% WACC).

Net income diverging from cash flow? FY2025 both are inflated (NI by the Hulu gain, CFO by Hulu cash/tax timing); on a clean basis CFO (~$28B) exceeds NI (~$13B) by the normal D&A gap — healthy, no red flag.

Risks & Downside

What would cause the stock to decline? Accelerating broadband sub and ARPU losses; a failed convergence/free-line monetization; a consumer recession hitting parks and advertising; a Peacock/streaming-loss relapse; value-destructive M&A (the Sky risk recurring); a permanent founder-control discount.

Catastrophic-loss risk? Very low — investment-grade balance sheet (2.3x leverage, 4.0% coupon, laddered), ~$12–13B FCF, diversified across connectivity/parks/content. No solvency risk.

Total-loss risk? Negligible. The realistic downside is a value trap (cheap stays cheap) and a ~20–30% drawdown in a bear scenario, not impairment of capital.

Recent News & Events

Has the business environment changed recently? Yes — materially: Versant spin (Jan 2026), Epic Universe opening (May 2025), the deliberate broadband go-to-market/pricing reset (2025), the NBA rights deal onset (2025-26), and a leadership reset (Cavanagh co-CEO, Croney to C&P). Competitively, fiber overlap crossed ~55% and Starlink emerged as a new broadband flank.

Significant acquisitions / accounting changes? Divestiture (Versant) and segment reorganization (2025) rather than acquisitions; no aggressive accounting changes.

Recent changes — new markets/facilities/management? New parks (Epic open; Frisco 2026; UK ~2031); DOCSIS 4.0 footprint completion; T-Mobile business MVNO added; co-CEO and C&P leadership changes. The news tape is otherwise quiet (Rosenblatt PT cut to $24, June 2026); short interest benign (~2.2% of float) — a neglected value name, not a contested short.

APPENDIX B — Source Appendix

As-of 2026-06-12. Public primary sources prioritized.

Primary — SEC Filings (EDGAR, CIK 0001166691)

1. Comcast FY2025 Form 10-K — filed 2026-02-03 (period ended 2025-12-31). Segment revenue/Adjusted EBITDA tables, broadband/wireless/video KPIs, Sky impairment history, debt schedule, capex. https://www.sec.gov/cgi-bin/browse-edgar?action=getcompany&CIK=0001166691&type=10-K
2. Comcast Q1-2026 Form 10-Q — filed 2026-04-23 (period ended 2026-03-31). Post-Versant segment view; broadband −65K, ARPU −3.1%, wireless 9.74M lines.
3. DEF 14A Proxy Statement — filed 2026-04-24. Dual-class voting (Class B 15 votes, Roberts 33⅓% non-dilutable), PSU metrics (ROIC + relative Adj-EPS + rTSR modifier), NEO compensation, say-on-pay 90%, board independence (82%).
4. Form 8-K — Versant spin-off — Jan 2026 (distribution 1 VSNT per 25 CMCSA; record date Dec 16, 2025). https://www.sec.gov/Archives/edgar/data/0001166691/000095010326000079/dp239108_8k.htm
5. Q4/FY2025 earnings release (8-K) — Jan 29, 2026. Full-year KPIs, broadband −711K, parks $1B+ quarterly EBITDA.
6. Form 4 corpus 2021–2026 — insider transactions; zero open-market purchases (code P); all routine M/F/A/S/G activity.
7. EDGAR XBRL company-concept API — Revenues, NetIncomeLoss, OperatingIncomeLoss, PaymentsForRepurchaseOfCommonStock, PaymentsToAcquirePropertyPlantAndEquipment, PaymentsOfDividends, GoodwillImpairmentLoss.

Primary — Transcripts (event documents)

8. Q1-2026 Earnings Call — Apr 23, 2026. Broadband stabilization framing, “Legendary February,” free-line conversion, parks +33%.
9. Q4-2025 Earnings Call — Jan 29, 2026. Parks $1B/qtr, segment reorg, go-to-market pivot.
10. Q3-2025 Earnings Call — Oct 30, 2025. Sentiment trough.
11. Versant Investor Day / Special Call — Dec 4, 2025. Spin rationale, RemainCo growth-driver framing.
12. Evercore Global TMT Conference — Jun 2, 2026. Peacock profitability confirmed for Q2-2026; “undervalued” framing.
13. MoffettNathanson Media/Internet/Communications Conference — May 14, 2026.

Secondary — Industry / Competitive

14. Leichtman Research Group — US broadband subscriber/net-add data (cable ~62.5% of base; ~41% of Q4-2025 gross adds). https://leichtmanresearch.com/research/
15. New Street Research via Fierce Network — FWA ceiling raised to ~32M subs; deceleration into 2026. https://www.fierce-network.com/broadband/big-3-now-have-room-32m-fwa-customers
16. Light Reading — Comcast DOCSIS 4.0 deployment progress (FDX amplifiers across footprint; symmetrical multi-gig). https://www.lightreading.com/cable-technology/comcast-s-docsis-4-0-deployment-now-covers-millions-of-homes
17. CNBC — Epic Universe ~$7B build (May 2025 opening); Sky $39B takeover (2018). https://www.cnbc.com/2025/05/22/epic-universe-opens-universal-orlando-florida.html
18. Variety / TheWrap — Versant spin close; Hulu $9.4B gain (Q2-2025); NBA 11-yr / ~$2.5B-yr rights deal. https://variety.com/2026/tv/news/comcast-completes-versant-spinoff-stock-public-company-1236623658/
19. Comcast Q3-2022 release — $8.6B Sky impairment.

Quantitative cross-checks (public market data; reconciled to filings)

20. Public market-data sources — price ~$24.50, market cap ~$87.5B, EV ~$173B, total debt ~$94.6B, short interest (~2.2% float), ownership (insiders ~1%, institutions ~89%), and own-history valuation percentiles (composite ~2.4th percentile). Reconciled to SEC filings, which remain primary.