CMCSA: Peacock turned profitable. Why did cash fall?

The 60-second read

Five things to know

1. Streaming is contributing. Peacock generated $261M adjusted EBITDA, its second profitable quarter; standalone cash flow is not disclosed.
2. The cash engine is weaker. Residential EBITDA declined $374M and parks EBITDA declined $160M year over year.
3. Definitions change the headline. Revenue fell 6.1% reported but grew 0.5% on the issuer pro forma perimeter; both must be labeled.
4. Buyback support paused. No Q3 program repurchases; the model assumes no automatic share-count shrinkage.
5. Cash value is near the price. Three cautious routes support about $23 against $22.84, with a $16–32 ordinary scenario band.

Three scenarios, one probability-weighted number

Analyst scenarios; weights express uncertainty, not observed probabilities.

Signposts: what would change our mind

Thresholds are analytical checkpoints, not guidance. Seasonal Q4 cash must be reconciled before interpreting a miss. Reporting day is not issuer-confirmed.

The price and the hurdle

Comcast rose 3.44% to $22.84 on October 22, the day it reported the September quarter. That is an observed close-to-close price change, not evidence that every investor interpreted the release the same way. The response follows a long decline: the saved provider quote places the stock below both its fifty-day and two-hundred-day averages. A positive session can therefore coexist with a market still demanding proof of stabilization. We anchor all valuation comparisons to that dated close, not to an unstated live price.

The interesting question is what cash the price requires from the businesses investors actually own. Provider market capitalization is about $80.84 billion. Add $79.77 billion of balance-sheet net debt and enterprise value is about $160.61 billion. At our assumed 4.9 times normalized adjusted EBITDA, that implies roughly $32.78 billion of sustainable annual EBITDA. The base case uses $33 billion. In other words, under this deliberately cautious framework, the stock already trades near the central operating outcome; a low headline earnings multiple alone does not create a fresh margin of safety.

The historical chart crosses changes in corporate perimeter. Versant was separated in January and Sky Germany was sold in May. Ordinary price returns in this packet are not reinvested total shareholder returns and do not add distributed securities or dividends. The nine-row reaction inventory retains unavailable entries rather than filling older dates with invented prices. The research is dated October 22 and captures a later provider retrieval; it does not claim that vendor consensus snapshots were archived before the earnings announcement.

What the quarter actually says

The release presents three-month 2026 results first, followed by the 2025 comparison; separate nine-month columns must not be confused with the quarter. Revenue was $29.298 billion, down 6.1% as reported. On the issuer's pro forma perimeter, excluding businesses that have left, revenue grew 0.5%. Both statements are true. The first describes the consolidated accounts; the second asks how the retained set compares with a reconstructed prior period. The adjustment does not turn weak broadband pricing into growth.

GAAP diluted earnings rose to $0.97, while adjusted EPS fell to $1.07. That opposite direction is the central reason not to summarize the quarter with a single beat percentage. Parent net income increased to $3.448 billion, but adjusted net income declined to $3.824 billion. The weighted diluted share count fell from 3.689 billion to 3.562 billion, helping per-share results relative to total profit. Because new common-stock repurchases stopped in the third quarter, the prior rate of share-count reduction should not be carried forward mechanically.

The vendor earnings snapshot shows adjusted EPS of $1.07 against a $0.986 estimate, a computed 8.52% difference; revenue is 0.26% above its saved estimate. Those are contextual vendor comparisons, not an independently reconstructed pre-announcement consensus. We do not treat a small revenue beat as a substitute for examining the economics underneath it. There is no numerical consolidated annual EPS or revenue guidance asserted in this research. Management's strategic commentary and our model assumptions remain clearly separate.

Where the improvement gets absorbed

Peacock generated $261 million of adjusted EBITDA, compared with a $217 million loss a year earlier. The $478 million swing is meaningful evidence that streaming can contribute rather than consume segment earnings. Revenue was approximately $1.8 billion and paid subscribers reached 48 million. These rounded revenue figures and an EBITDA measure are not audited standalone net income, free cash flow or a separately valued security. Content spending and distribution arrangements still determine how much of that success reaches shareholders.

Residential Connectivity and Platforms lost $374 million of adjusted EBITDA year over year. Theme Parks lost another $160 million. Together, those two declines exceed Peacock's improvement, although this is a diagnostic comparison across disclosed businesses, not a consolidated earnings bridge: Peacock is already inside Media, and intersegment eliminations and other units also matter. Business Services grew, with $1.478 billion of EBITDA, while Studios benefited from a strong theatrical slate. Diversification helps, but it cannot make the mature cash engine irrelevant.

The most useful operating distinction is between wireless service growth and the total convergence wallet. Wireless service revenue rose to $1.050 billion from $915 million, helped by line growth and conversion of free promotions to paid relationships. Domestic broadband revenue fell to $6.241 billion from $6.550 billion. Combined domestic convergence revenue declined to $7.292 billion from $7.466 billion. Counting mobile lines without checking the combined revenue pool would miss the tradeoff. We want retained broadband households that generate more economic cash, not merely a higher number of products in a bundle.

Residential broadband losses were 144,000 versus 91,000 in the comparison quarter. These are residential figures; they must not be mixed with older total domestic broadband disclosures that included business customers. Premium product adoption can support retention, but cheaper introductory offers and customer equipment investment can delay the financial payoff. Segment revenue totals include internal transactions, so adding them without eliminations overstates consolidated sales.

Follow the profit to common shareholders

The income statement begins with $5.320 billion of operating profit. Interest expense uses $962 million, while investment and other income contributes $88 million, producing pretax income of $4.446 billion. Tax expense of $1.084 billion leaves consolidated net income of $3.362 billion. Noncontrolling interests absorb a loss of $87 million, so parent-attributable income is higher than consolidated net income. The published rounded totals differ by one million in the final reconciliation; this is a disclosed rounding convention, not an unexplained economic gain.

The adjusted net-income bridge is narrower than a casual comparison of EBITDA and earnings would suggest. After-tax acquisition amortization adds $402 million, investment adjustments subtract $95 million, transaction costs add $19 million, transaction-related costs add $4 million, and other adjustments add $46 million. These take reported parent profit to adjusted net income of $3.824 billion. The economic question is which exclusions recur, not whether the adjusted label is inherently good or bad. Software amortization and acquired rights used in parks are not all removed simply because they are intangible.

Comcast separately adjusts EBITDA for $22 million of transaction costs, $4 million of transaction-related costs and $61 million of other items. The separation may have a finite project scope, but its cash expenses are real. Some share-based compensation also appears in the description of adjustments. Our economic cash scenarios retain a recurring compensation cost rather than assuming every stock award can be added back forever. The nine-month comparison is especially distorted by prior-year investment gains, so neither year-to-date GAAP EPS nor a single trailing earnings multiple is our principal valuation anchor.

Nine checks on earnings quality

Cash conversion is strong in an absolute sense: operating cash flow of $7.902 billion is 2.29 times parent net income. Yet conversion does not establish that the current cash rate is sustainable. Depreciation is large, reinvestment is substantial, and cash taxes and content payments move between periods. The operating-cash decline explains nearly all of the year-over-year free-cash decline. Lower tangible capital expenditure and lower intangible cash spending only partly offset it. Treating the gap as a harmless accounting presentation would miss a real reduction in cash generated.

Receivable days are approximately 42.5 using ending net receivables divided by quarterly revenue and multiplied by 92 days. This is a rough diagnostic, not a full average-balance collection cycle: business mix, equipment installment plans and the separation perimeter affect it. The current report separately identifies noncurrent receivables, including installments. We do not manufacture inventory days from a provider's zero inventory field. Film and television costs also have different economic meaning from ordinary industrial inventory, and an inventory-turn formula would obscure that difference.

The effective quarterly tax rate is 24.4%. Annual cash taxes can be much lower than income-tax expense, as the 2025 annual report describes refunds, deductions and timing. That distinction helps explain why the extraordinary 2025 free-cash-flow level should not automatically become the permanent run rate. Guidance quality is marked ungraded because this packet does not establish a numerical consolidated annual guide-and-deliver series. Management's disclosure-control conclusion is relevant evidence, but it is not a guarantee of future operating execution.

These nine checks are a repeatable framework, not a fraud score or a substitute for reading the filing. Mixed results are expected for a capital-intensive conglomerate. The most important warning is deterioration in cash generation while the legacy customer engine is being repriced. We would become more constructive if retention improves without another margin reset and the content gains remain cash-producing after their associated programming commitments.

Reinvestment and the debt claim

Comcast defines free cash flow after both tangible capital expenditures and cash paid for capitalized software and other intangibles. The quarter therefore reconciles as $7.902 billion of operating cash less $3.058 billion of capex less $633 million of intangible investment, leaving $4.211 billion. A generic provider calculation returned about $4.845 billion because its treatment omitted the intangible deduction and had a one-million-dollar operating-cash difference. That larger number is not used as Comcast's free cash flow anywhere in our model or script.

Balance-sheet cash was $7.596 billion and carrying debt was $87.366 billion, producing $79.770 billion of net debt. Cash and restricted cash in the cash-flow statement is a different measure; restricted balances are not silently treated as freely distributable cash. Net debt per actual quarter-end Class A plus Class B common share is about $22.48. That is a debt-allocation illustration, not a claim that shareholders must personally pay a bill of that amount. It explains why a modest shift in enterprise value can have an outsized effect on the residual equity.

A subsidiary separation does not automatically erase consolidated economic claims. Future debt allocation, retained guarantees, standalone costs, tax treatment and cash transferred between entities all matter. The proposed NBCUniversal and Sky transaction remains conditional. Our base valuation assesses the combined economics and makes no unsupported assumption that both successor equities receive premium multiples. It also avoids adding the same cash flows twice through a spin-off bonus on top of a whole-company valuation.

Dividend payments continued, while the common-stock repurchase program was suspended from the start of Q3. A small cash-flow entry for repurchases under employee plans does not contradict the disclosed absence of program repurchases. Cash can instead repay debt or fund operating investment. That may strengthen the business over time, but investors should no longer rely on the earlier buyback pace to cushion every weak operating quarter. We hold the prospective diluted share count at 3.550 billion in the model rather than inventing future shrinkage.

Three ways to earn a $23 value

Our call is HOLD with moderate conviction, three out of five, and high uncertainty. Central fair value is $23, close to the $22.84 reference price. The ordinary scenario band is $16 to $32 and the probability-weighted value is $22.70. These are analyst judgments, not price predictions or issuer targets. We require a twenty-percent discount to central value before a valuation-only BUY, corresponding to $18.40 if the business assumptions remain intact. A falling price caused by deteriorating fundamentals does not automatically qualify.

The enterprise route assumes normalized annual adjusted EBITDA of $33 billion and a 4.9 times enterprise multiple. That produces $161.7 billion of enterprise value; subtract $79.77 billion net debt and divide by 3.550 billion prospective diluted shares to reach $23.08. We deliberately do not multiply the strongest current quarter by four and call it guidance. The normalized figure allows for weaker connectivity economics and portfolio changes. The chosen multiple reflects competition, capital needs and execution risk; it is not justified by a claim that a breakup must occur.

The earnings route uses $3.20 of normalized economic EPS at 7.2 times, producing $23.04. This is a cross-check, with recurring compensation and financing costs treated as economic expenses. It is not a forecast of issuer-adjusted EPS and should not be compared with that measure without accounting for the definition. The equity cash-flow route uses $12.0, $11.5, $11.2, $11.0 and $11.0 billion over five years, a fourteen-percent required equity return, and zero terminal growth. The resulting value is $22.53 per share. Because these are cash flows after interest, debt is not subtracted again.

The initial economic cash assumption starts from $13.5 billion of normalized issuer-style free cash flow and deducts $1.5 billion for recurring equity compensation. It is below reported trailing free cash flow of $17.085 billion because favorable tax timing, earlier business perimeter and uncertain cash conversion should not be capitalized indefinitely. Roughly 51% of the discounted equity value still comes from the terminal component. The sensitivity grid makes the model's fragility visible: small multiple and earnings changes matter when debt is almost as large as the equity market value.

What Wall Street is pricing

The canonical Street snapshot reports an aggregate Buy classification with a $26.95 consensus target, versus our $23 central value. Its ratings inventory contains 34 Buy, 24 Hold and three Sell observations. Counts are vendor observations, not proof of sixty-one independent analysts issuing fresh reports after this quarter. A target average combines dates, business assumptions and methodologies. It is useful context, but we do not convert it into a probability distribution or silently treat its spread as a valuation model.

There is a fresh target-news item on October 22 showing Goldman Sachs reducing its target to $23 from $24. That is after the print; the saved grades feed does not independently confirm a matching new rating, so no post-results rating is invented. Bernstein's October 13 target of $23 and Scotiabank's October 2 target of $27.50 predate the release. Barclays' September 29 target is $24. The sequence illustrates caution already in the market before Q3 rather than a unanimous overnight reassessment.

Annual vendor estimates put 2026 revenue around $119.95 billion and EPS around $3.45, followed by 2027 revenue around $117.38 billion and EPS around $3.55. These are a dated provider snapshot and may mix analyst definitions or responses. They are not numerical management guidance, and they are not a clean revisions series because we have not reconstructed identical analyst panels at two dates. Our lower normalized economic EPS reflects a different purpose: a through-cycle valuation check after recognizing recurring costs, rather than a precise next-quarter adjusted-EPS forecast.

The disagreement is therefore specific. The Street average appears to allow more upside than our cash-and-enterprise cross-checks do. We would narrow that gap if the company demonstrates that the customer-offer reset produces a durable combined revenue base without further margin sacrifice. We would not raise fair value merely because a familiar bank changes its target or because a separate ticker is expected in the future.

Grade actions against outcomes

Management deserves credit for getting Peacock into positive adjusted EBITDA and converting some promotional wireless lines into paid relationships. Those outcomes are observable. The unanswered issue is whether the residential strategy can preserve customer economics while competing against fiber, fixed wireless and other offers. Higher satisfaction, premium plan adoption and wider product penetration can be leading indicators, but investors ultimately need margins and cash. We treat management's strategic language as a hypothesis that future quarters can test.

The capital-allocation score is mixed. Suspending buybacks while preparing a major separation preserves financing flexibility and can be sensible. It also removes a repeated source of per-share support at a time when adjusted net income is declining. We do not assume that buybacks restart on a particular date or that the two companies inherit a predetermined debt split. The annual report establishes a concentrated governance structure: Class B carries one-third of aggregate voting power, and Brian Roberts controls that block. Public Class A shareholders do not have equal influence per dollar invested.

The 2025 annual report lists Deloitte & Touche as auditor and describes effective internal control over financial reporting. It also reports approximately 179,000 full-time-equivalent employees, including full- and part-time workers, at year-end. The provider calls that a full-time employee count, so we retain the issuer's more precise definition. Investments and variable-interest arrangements, including Atairos, require their own accounting attention; no unsupported blanket absence of VIE exposure or related-party risk is claimed.

Our next management questions are concrete. How much wireless revenue comes from converting free lines rather than additional households? What broadband loss level is compatible with stable residential EBITDA? How much of Peacock's benefit survives its sports-payment cycle? What standalone and transaction costs remain to be paid? How will debt and guarantees be divided if the separation closes? Answering these matters more than repeating that the stock is statistically cheap.

What can break either case

The largest operating risk is a residential price-and-retention loop: stronger offers can retain customers but compress revenue per relationship, while weaker offers can accelerate churn. If the company needs both more equipment spending and deeper promotions, EBITDA may overstate the cash left after maintaining the franchise. We rank this high likelihood and high impact because current broadband revenue and residential margins already show pressure. It is not a prediction that cable networks become worthless.

A second risk is assuming streaming profitability is smooth. Sports rights, film releases, advertising and distribution renewals create uneven comparisons. Peacock's improvement is real, but two profitable quarters do not eliminate the economic risk of content commitments. Theme Parks adds cyclical exposure to discretionary travel and promotional intensity. Debt makes all these operating uncertainties more important to equity value. The ordinary bear scenario is not a stress test for every possible recession, litigation outcome or failed transaction.

The strongest bull argument is that investors are paying a depressed valuation just before wireless monetization and streaming profitability compound. Business Services can provide resilience, parks can recover and a clearer corporate structure could improve accountability. We accept those mechanisms as plausible. The strongest bear argument is that the legacy engine is funding improvements elsewhere without enough consolidated cash growth, while buyback suspension exposes the underlying profit trend. Our HOLD gives neither side a free assumption: the price is near cautious central value, and the signposts decide whether that value should move.

Separation carries tax, regulatory, financing and execution risk. It can be delayed or fail to achieve its intended benefits. It can also succeed while initially increasing standalone costs. Portfolio comparison adjustments are therefore an analytical necessity, not a reason to ignore reported declines or declare all historical data unusable. We keep the continuing listed parent eligible for coverage while making the uncertainty explicit in the valuation and monitoring plan.

The next decision points

The next earnings package should answer whether residential margin erosion is slowing and whether customer losses improve on the current 144,000 quarterly residential baseline. We deliberately pair the two: a better customer count purchased with another large margin decline is not an unqualified success. A third profitable Peacock quarter would strengthen confidence in the direction, while disclosures on recurring programming and transaction costs would help separate operating durability from timing. None of these requires waiting for a stock-price milestone.

A separation filing, board decision or financing disclosure can change the model when it supplies concrete allocation terms. Until then, there is no invented closing date or debt split. The provider calendar currently points to a February 2027 results window, but this packet labels the next Q4 review as an expected monitoring window, not a confirmed issuer event. We use February 15 as an analytical check date so the signposts remain accountable even if the company chooses a different reporting day.

We would upgrade the operating thesis after a combination of stronger residential retention, stable-or-better margin, sustained positive streaming EBITDA and cash generation that covers required investment. We would lower central value if convergence revenue keeps declining while pricing investment intensifies, or if new separation terms transfer substantially more economic claims to the retained business than assumed. An eventual buyback restart could help per-share value, but it must compete with debt reduction and investment on expected return rather than serve as a headline catalyst.

The prior channel coverage emphasized an earnings beat and a cheap-stock/spin-off case. This review intentionally tests the new cash evidence instead. No earlier packet signpost set was found in the current canonical prior-quarter directory, so we do not fabricate a historical grading record or imply that the $23 value was previously forecast. These dated thresholds establish the comparison for the next research packet.

Data, definitions and audit trail

The research dataset preserves nine original earnings releases from Q3 2024 through Q3 2026, their full table inventories, detailed provider income statements, balance sheets and cash-flow histories, and three years of annual cash data checked against the 2025 Form 10-K. Current issuer disclosures govern where a vendor differs. Older reported results retain their original corporate perimeter; the packet does not splice them into a synthetic continuing-company history. Non-GAAP measures retain the issuer's name and definition rather than being relabeled as GAAP operating profit or distributable cash.

All derived margins, changes, per-share debt illustrations, enterprise arithmetic, scenario weights and discounted values are computed in the retained Python calculation file. Forecast cash flows, multiples and probabilities are explicitly analyst assumptions. Debt in the enterprise route uses carrying amounts and unrestricted cash; it does not assign an unsupported premium to investments. The equity DCF already includes financing costs and therefore does not subtract net debt a second time. Future modeled compensation is economic cash dilution, not a claim that all awards require immediate cash settlement.

Source retrievals are timestamped and hash-bound in the episode's retained research directory. Comcast's release was public before the October 22 U.S. market opening; SEC acceptance establishes an upper bound of 6:05:59 a.m. Chicago time. That is not represented as an exact release minute, and the conference-call time is a separate event. No call transcript quotation or unverified numerical annual guidance is included. Historical employee counts and inventory days that were not established are marked unavailable. Provider consensus estimates can change and should be compared with later snapshots before drawing revision conclusions.

Adjusted EBITDA means earnings before interest, taxes, depreciation, amortization and specified issuer adjustments. It is useful for comparing operating capacity, but it excludes major economic claims. Free cash flow here means Comcast's reconciled cash after tangible capital expenditures and capitalized software/intangibles. Pro forma describes a reconstructed historical perimeter, not a forecast. The fair-value range describes selected ordinary scenarios; it does not cap losses or guarantee an investment return. Every rating in this packet is a conditional research conclusion, subject to the evidence and limitations stated above.