Actionable insights — Comcast: The Great Unbundling

1. Ask whether "over-spending" is a mistake or a business waiting to be sold

1. When a company is being punished for a massive CapEx line, don't stop at "they over-built" — ask whether the excess asset can be rented or resold, converting a cost into a revenue stream.
2. Look for the tell that management sees capacity worth selling: a public comment ("on the table"), inbound demand ("people ask to buy it every week"), or an actual product launch — that reframes the spend from waste to option value.
3. Anchor it to precedent: the same excess-capacity-becomes-a-product arc has played out before (Amazon's spare compute → AWS), so weight the base rate that a scaled owner of a scarce asset can monetize it.

• Whether the new line actually books revenue and at what margin; a management comment shifting a cost narrative to an option narrative; and which incumbents/customers it disrupts (below).

2. Trace who gets hurt when a big customer becomes a supplier

1. When a scaled buyer announces it will start selling what it used to buy, map its former suppliers/vendors — they face a double hit: losing the customer and gaining a competitor at once.
2. Rank the exposure by how concentrated each vendor's revenue was on that buyer — the more dependent, the sharper the sell-off.
3. Separate the share-price reaction (immediate, sentiment-driven) from the fundamental question (does the buyer actually have the will and cost structure to compete for third-party business).

• The disclosed customer-concentration on the newly-competing buyer; whether the sell-off overshoots (the buyer may never seriously chase third-party rental); pricing pressure as new supply lands.

3. Value a breakup on sum-of-the-parts, then check what the upside actually requires

1. For a conglomerate splitting up, value each piece on its own — a stable cash-throwing "utility" and a lumpy hit-driven business deserve different multiples, and bundling them tends to depress the blended valuation.
2. Compare the sum-of-the-parts estimate to the current price: if SOTP is only modestly above today's price, the split alone isn't the catalyst — the extra upside must come from something else (usually M&A that the newly-independent equity currencies enable).
3. Discount management's denials: a company that says a split is "absolutely not" a prelude to deals has just handed each half its own stock, board, and focus — the classic pre-condition for dealmaking.

• The SOTP gap vs the current price; whether each spun entity gets a clean equity currency and independent board; the segment divergence that justified the split (see #4); tax-free-separation timing and any stake the parent retains to sell down.

4. Use segment divergence — not the blended number — to justify (or doubt) a split

1. Break the conglomerate into its reported segments and compare growth trajectories: a flat, declining annuity bundled with a volatile high-growth engine is the textbook case for separation.
2. Strip out non-recurring boosts before judging the "growth" segment — a spike driven by one-off events isn't a run-rate.
3. Ask which investor base each segment attracts; if they're incompatible (yield-seekers vs growth buyers), the bundle structurally under-earns its parts.

• Whether the high-growth segment's spike is event-driven (non-recurring); each segment's structural threats (fixed-wireless/Starlink on connectivity); and management's own stated rationale.

5. Strip one-time items before believing a "blowout" — the margin-mirage test

1. When reported EPS or gross margin spikes, hunt the footnotes for a one-time item (a tariff refund, legal settlement, insurance recovery) and quantify its EPS and basis-point contribution.
2. Recompute the "clean" number: subtract the one-off, then compare that against consensus and against the prior trend — a headline blowout can shrink to a modest beat.
3. Confirm with the top line and constant-currency revenue: a genuine recovery shows up in growing sales, not just a flattered margin.

• The size of any one-time item vs the headline beat; underlying (ex-item) gross margin vs the reported figure; constant-currency revenue direction; whether real green shoots (a segment or category) back the "recovery" claim before paying up (NKE ~28x forward, still "show me").