Crown Castle Inc. — 5.1/10 — $76.74

Crown Castle is now a pure-play US tower REIT. The fiber and small-cell business — built over a decade for well north of $10B of invested capital, publicly defended, then written down — closed its sale on 2026-05-01 for $8.5B headline / $8.4B net proceeds. What remains is ~40,000 macro towers, 100% domestic, with site rental at 95.9% of revenue and a small, shrinking services book at 4.1%.

The structural case is real. A macro tower is a zoning-and-siting monopoly at a fixed coordinate: municipal permitting, environmental review, FAA clearance and ground-lease control make an equivalent structure inside a carrier's search ring effectively un-replicable at economic cost, which is why AT&T's master lease lets it add radios at 30-40% below new-build economics. CCI holds ~37% of the US third-party tower-leasing revenue pool and is one of three firms controlling just under 70% of privately owned US sites, sitting on $26B of contracted receivables against a ~$4.0B revenue base.

What that moat is not producing is growth. The US tower TAM compounds at only ~3.4%, and CCI's own realized theme growth is negative — organic contribution to site rental revenue has gone +4.4% (FY2024) to -0.3% (FY2025) to -2.5% (26Q1) to -1.8% (26Q2). Tower count has been static at 40,000 for three years, and revenue per tower is falling from ~$107K LQA in 24Q4 to ~$97K today. Sprint and DISH decommissioning churn ($240M combined FY2026 headwind) more than consumes a healthy +3.9% underlying organic billings number.

The cash-flow improvement is financial engineering, not operations. The headline "AFFO +9.9%" is an easy-comp artifact off an outlier-low 2Q25 of $444M; on a clean half-year basis 1H25 $923M became 1H26 $934M, or +1.2%. The real driver was $8.4B of proceeds funding $7.2B of debt repayment and a $1B buyback — a $35M drop in interest expense plus $14M of interest income that management itself flagged as non-recurring. Durable, but a one-time step-function.

PARTIAL PASS — 1 NO. No cap applied (a single NO means score normally, with the gap flagged prominently; caps apply only at 2 NOs and 3 NOs).

The gap, stated prominently. The single failing gate is the one that matters most for an 18-36 month thesis. Every forward claim here — that 2026 is the organic-growth trough, that ~200bps of EBITDA margin is recoverable from ground-lease buyouts, that edge compute monetizes the 40,000-site footprint at zero capex — is being made by a pairing that has been together ten months at a company whose previous strategic conviction was built over a decade, defended publicly, then sold for roughly half of the capital invested. The in-year execution record is genuinely strong. What does not exist is a multi-year record. Underwrite the execution; do not yet underwrite the reacceleration.

PM review caveat. The PM pass judged 5.1 "roughly right for the wrong reasons — an over-harsh Financial Trends offsetting an over-generous Thematic," with an honest range after corrections of 4.6-5.0. The corrections it demanded have been applied to every page here: the AFFO growth claim is restated on a 1H basis, the oligopoly denominator is published with its sensitivity, the margin trend uses YoY-compliant comparisons only, the sentiment divergence is re-cut on P/AFFO, the enterprise value and leverage figures are the post-paydown ones, and the disputed tenants-per-tower claim has been removed (see Thematics).

Crown Castle is a real oligopolist trapped in a theme that is not growing, run by a team with ten months of history, priced at a premium to two better assets. 5.1/10 — average, and average is not a reason to own something.

The first principle is that leaders in a market remain leaders and you should not settle for the second, let alone the third. On asset quality CCI is the #3 of three: worse ground economics (a self-admitted ~11-point land-ownership gap), worse customer concentration (89% from three carriers, up from 73% a year earlier), worse growth, higher leverage, and a higher multiple. The trajectory principle cuts the same way — a business decelerating at -4.9% with compressing GAAP margins is not rescued by a balance-sheet event.

What holds the score at 5.1 rather than lower is that the moat is authentic, cash flow is positive and growing on a like-for-like basis, the share count is genuinely shrinking, China exposure is zero, the regulatory backdrop is a net tailwind, and the catalyst calendar is unusually well-dated.

This is a watch item, not a buy. The thesis becomes ownable on evidence, not narrative. Absent that evidence, it is a high-yield bond proxy with a shrinking annuity and a premium multiple.

Key question: does the 3.9% underlying organic billings growth actually reaccelerate through the Sprint/DISH churn in 2027 as management promises — or does a 92.6% AFFO payout, 6.33x leverage and a 2028 AT&T renewal with the tenant holding the leverage mean the annuity keeps shrinking faster than the self-help can offset?

• Whether 2026 actually is the organic trough. Management has now said it on four consecutive calls, with FY26 organic nudged 3.3% to 3.4% and the contracted share of that growth up from ~80% to over 90%. A credible multi-year FY2027 guide that prints the promised reacceleration is the single thing that resolves this thesis.
• DISH bankruptcy resolution. Bid deadline Aug 10, 2026, auction Aug 12. CCI sits on the unsecured creditors' committee pressing a $3.5B claim against a bankruptcy-remote $2.4B FCC-mandated escrow, with DISH litigating an ~85% statutory lease-cap haircut. Hold it as a lottery ticket, not an asset.
• The ~200bps of EBITDA margin from ground-lease buyouts. A quantified, closable self-help lever against a measured 11-point land-ownership gap to AMT and SBAC. Q2 already delivered $15M of in-year cost take-out on top of the $65M restructuring.
• AFFO payout and retained cash. At 92.6% of FY26E AFFO, roughly $164M is retained against $200M of discretionary capex — there is no organic deleveraging and no organic buyback capacity now that divestiture proceeds are spent.
• Insider buying. The clearest missing confirmation. The company bought back $1B of stock at $88.66 while the CEO has not purchased a share personally with the stock at $76.74.
• The 2028 AT&T renewal. ~$774M of annualized rent struck under 2013 sale-leaseback terms. Asked directly, management offered only "win-win outcomes."
• Edge compute moving from trial to quantified revenue. Management calls it "gaining momentum" and zero-capex; the street carries zero, and management has quantified nothing.