American Tower

Phase A — Understand the business

Lens 1 · Company Overview

American Tower is one of the largest global REITs and the leading independent owner/operator/developer of multitenant communications real estate. The model is dead simple and extraordinarily durable: AMT owns the steel (towers) and the ground rights; wireless carriers bolt their radios onto it under long-term, escalating, non-cancellable leases; each incremental tenant on an existing tower drops almost entirely to gross margin. It is a tollbooth on mobile data traffic.

Scale. As of 2025-12-31: 149,686 communications sites — 42,224 U.S. & Canada, 27,857 Africa & APAC, 32,524 Europe, 47,081 Latin America — plus 30 CoreSite data centers across 11 U.S. markets (3,773k net rentable sq ft). 4,866 full-time employees across 20+ countries. HQ Boston, MA; incorporated Delaware; REIT for U.S. federal tax.

Six reportable segments: U.S. & Canada property, Africa & APAC property, Europe property, Latin America property, Data Centers, and Services. Property operations = 97% of revenue (2025); Services (US tower construction/zoning/permitting) = 3%.

Contract structure — the moat in the fine print. Tenant leases run 5–10 year initial non-cancellable terms with multiple renewals, escalating ~3% fixed in the US or by an inflation index internationally. Churn has historically been benign (~2% of tenant billings in FY2025) — suitable alternative sites rarely exist and repositioning a radio degrades network quality, so tenants renew. AMT discloses >$54B of contracted, non-cancellable future tenant lease revenue as of 2025-12-31 (before straight-line accounting) — a ~5x-revenue visible backlog.

Customers / suppliers / competitors: see Lens 2 (chain) and Lens 3 (moat). Top tenants are the mega-carriers (AT&T, T-Mobile, Verizon in the US). Suppliers are minimal — ground-lease landlords, steel/construction, and power utilities (the binding constraint for the data-center arm). Competitors: Crown Castle (CCI), SBA Communications (SBAC), Cellnex, Vertical Bridge, Telesites.

Lens 2 · Supply Chain

A tower REIT's "supply chain" is land rights upstream → the tower asset → carrier tenant downstream, plus power as a swing input. Named stakeholders:

Upstream (inputs):

• Ground landlords — AMT leases (and increasingly buys) the land under its towers. Ground rent is the largest direct operating cost; "Property Interests" portfolios in the US/Canada are AMT buying its own land to capture the spread.
• Power utilities — pass-through fuel/power costs internationally (Africa especially); for CoreSite, utility power is the chokepoint: utilities now impose multi-year interconnect queues and require operators to pre-pay for a high share of requested capacity. This is the single hardest constraint on data-center growth.
• Construction / steel — site development for ~2,230 new sites built/acquired in 2025.

The asset: 149,686 towers + 30 data centers. AMT also runs DAS networks, rooftop management, fiber (international), and shared power/generators (PaaS in Africa).

Downstream (tenants) — concentrated by segment:

• U.S. & Canada: AT&T + T-Mobile + Verizon = 85% of segment revenue.
• Africa & APAC: Bharti Airtel + MTN = 81%.
• Europe: Telefónica = 70% (single-name dependency — a chokepoint).
• Latin America: América Móvil + AT&T + Telefónica + TIM = 71%.
• Data Centers (CoreSite): enterprises, network operators, cloud providers, and "neo-clouds" (GPU-as-a-Service) — diversified, interconnection-dense.

Chokepoints / single-source dependencies: (1) Carrier oligopoly — three US carriers are 85% of the largest segment; (2) Telefónica is 70% of Europe; (3) utility power queues gate CoreSite; (4) FX — most international rent collected in local currency, swinging reported revenue (see Lens 5).

Lens 3 · Competitive Advantages (moats)

The moat is real and multi-layered, but it is a "high-quality utility" moat, not a hyper-growth moat.

1. Switching costs / irreplaceability (strongest). A carrier cannot easily move a radio — alternative sites with the right height, zoning, and coverage geometry often don't exist, and repositioning degrades the network. This is why churn is ~2% and renewals are near-automatic.
2. Operating leverage / incremental-tenant economics. Adding a second/third tenant to an existing tower costs almost nothing — "the substantial majority of incremental revenue flows through to gross margin." US & Canada segment gross margin runs ~84% ($1,055.3M on $1,261.6M, Q1 2026).
3. Scale + irreplaceable footprint. 149,686 sites is a portfolio no new entrant can replicate; ~108,000 are international, positioned to ride the same 4G→5G densification curve the US already climbed.
4. Contracted backlog + escalators. >$54B non-cancellable future revenue with ~3% US escalators / inflation-linked international gives built-in low-single-digit organic growth before any new leasing.
5. CoreSite interconnection density (the option). Data centers at network-convergence points become interconnection hubs — a network-effects moat that hyperscale shells lack, and well-suited to hybrid/AI-inference workloads that want to sit next to cloud on-ramps.

Bargaining power: AMT holds power over tenants once equipment is installed (switching cost), but the carrier oligopoly holds power at the negotiating table — three buyers, and master-lease agreements compress pricing in exchange for volume/speed. Against landlords, AMT increasingly removes the counterparty by buying the land. Against utilities (for CoreSite), AMT is a price-taker in a constrained queue.

Lens 4 · Segments

** — all segment figures below are filing-sourced.**

FY2025 revenue by segment (vs FY2024):

Q1 2026 revenue by segment (vs Q1 2025):

What the trends say:

• U.S. & Canada is decelerating into a DISH air-pocket. FY2025 segment revenue was flat — +$210M organic tenant billings growth was almost entirely offset by a −$209M drop in "other revenue" (−$175.8M of it straight-line accounting rolling off). Q1 2026 went outright negative (−3%) as DISH moved fully into churn (Jan 1, 2026). Underlying colocation/amendment demand is still healthy (+$34.0M colos in Q1'26), but reported optics are ugly.
• International is the reported-growth engine — but FX-inflated. Europe +22% and LatAm +20% in Q1'26 are heavily FX-aided (Euro, Brazilian Real, Mexican Peso translation). LatAm's underlying tenant billings actually fell $5.8M (Brazil churn); the +20% is FX + straight-line/settlements. Read international growth on an FX-neutral, organic basis, not the headline.
• Data Centers is the cleanest growth story. +14% FY2025, +18% Q1'26, driven by new lease commencements, power revenue, interconnection, and renewal mark-ups — genuine demand, ~61% gross margin (lower than towers because power is a cost pass-through). This is the segment whose growth rate the market actually cares about.

Geography: US is ~49% of revenue and falling as a mix (was 53% in 2023). International is more diversified but lower-margin and FX-exposed; AMT has been actively shrinking the riskiest international (India, Australia, NZ, South Africa fiber, Mexico fiber, Poland all exited 2023–25).

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest print: Q1 2026, filed 2026-04-28)

The headline beat, but the story is "underlying 4-5% organic dressed as ~1% reported."

Q1 2026 actuals:

• Total revenue $2,737.5M, +6.8% YoY ($2,562.8M). Property $2,669.9M; Services $67.6M.
• Operating income $1,239.2M (down slightly YoY — prior year had a $55.8M other-operating gain).
• Net income $878.5M (+76%), net income to common $859.5M; diluted EPS $1.84 (vs $1.04). The +76% is mostly an FX swing — Q1'25 carried a $345.7M FX loss; Q1'26 a $68.1M FX gain. Strip FX and the operating story is far flatter.
• Adjusted EBITDA $1,835.2M (+5%).
• Nareit FFO $1,278.9M (+57%) — distorted; the +57% is the prior-year FX-loss base, not operations.
• AFFO (common) $1,324.1M (+2.6%); AFFO/share ~$2.84. On an FX-neutral basis, normalized for one-time DISH churn and ex-refinancing, AFFO/share grew ~4%.

Drivers: Data Centers (+18%) and international tenant billings carried the quarter; U.S. & Canada (−3%) was the drag (DISH churn + straight-line roll-off). ~5% Q1'26 churn (vs ~2% normalized) is entirely the DISH effect.

Guidance — raised. Management raised the FY2026 AFFO/share midpoint by $0.12 to ~$10.99 (range ~$10.90–$11.07), and lifted property revenue, Adjusted EBITDA, and net income midpoints (net-income-to-common midpoint +18.8% YoY). FY2026 property revenue midpoint ~$10.66B (+3.4%). But guided organic tenant billings growth is only ~1% (≈4% ex-DISH); U.S. & Canada ~0.5% (≈4.5% ex-DISH); LatAm ~−3% on consolidation churn.

Balance-sheet flags: Cash from ops $1,400.6M (+8%); capex $449.5M; net debt ~$35B (total long-term obligations $37.3B incl. current; cash $1.6B). Current portion of long-term obligations jumped to $6,119M (from $3,388M) — a 2026 maturity wall to refinance at higher coupons (new 2025 notes priced 4.7–5.35%). Accounts receivable up to $718.9M (from $650.3M) — partly the DISH/AT&T Mexico disputes.

Market reaction: Despite the AFFO raise, the stock initially fell on "weak 2026 revenue guidance" — the market fixated on the ~1% reported organic line, not the ex-DISH ~4%. Stock ~$184 mid-June 2026.

Unusual vs own history: The flat-to-negative U.S. organic line is the most un-AMT-like datapoint in years; it is explained (DISH, straight-line) but it is why the multiple has compressed.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on disk (transcripts=0); this lens is ``.

Management tone across the last several calls has shifted from defensive (2022–23: rates, India, FX) to "quality-of-earnings + reset" (2025–26). On the Q1'26 call, leadership framed 2026 as a deliberate reset year — absorbing DISH and straight-line headwinds — while pointing to 4–5% FX-neutral organic underneath and a multi-year reacceleration into 2027+. The recurring phrases now: "disciplined capital allocation," "developed markets," "data center momentum / AI inference," "FX-neutral organic," "quality of earnings." The things they stopped saying: aggressive emerging-market expansion, India.

The tone is credible but carries a clear "trust us, 2027 reaccelerates" ask — the central debate (Lens 12).

Lens 7 · Comps

Tower/infra-REIT peer set. Multiples are ``, dated; where a clean figure isn't sourced I mark n/a.

Read: AMT trades at a discount to CCI (~17.8x vs ~23x EV/EBITDA) despite a more diversified, higher-quality, growing book — CCI's premium reflects its 100%-US, post-restructuring "clean tower" story and a higher payout, not better assets. AMT screens fair-to-slightly-cheap for the quality, with the data-center optionality arguably not in the multiple. P/AFFO ~16.7x on 2026E is below AMT's own historical ~20–24x.

Lens 8 · Stock-Price Catalysts (last ~5 years; ``)

What actually moves AMT:

• 2022 — rate shock (the big one): AMT de-rated hard as the 10Y spiked; ~72% of the negative impact that year was higher financing costs, ~17% India, ~11% FX; operating margin fell 38%→30%. Lesson: AMT trades like a long-duration bond proxy — rates are the dominant macro driver.
• 2024 — India exit: Sold 100% of India ops to Brookfield's Data Infrastructure Trust, ~₹210B / $2.5B, closed Sept 2024 — removed the single biggest emerging-market credit overhang (Vodafone Idea). Re-rating positive.
• 2024–26 — DISH saga: DISH default (notice Sept 2025; in default Jan 2026) → moved fully to churn 2026 → the dominant negative 2026 optics; paradoxically a de-risking once priced (DISH was ~2% of property revenue, now gone). Wall Street re-rated bullish after the DISH overhang clarified.
• CoreSite / AI: every incremental AI-inference / interconnection datapoint is now an up-catalyst; the data-center segment is the re-rating lever.
• FX & straight-line: quarter-to-quarter the reported numbers swing on currency and non-cash straight-line — these create noise, not signal.

Pattern: AMT reacts to (1) rates (macro, dominant), (2) single-customer credit events (DISH, VIL), (3) AI/data-center demand, and (4) FX. It is less sensitive to broad-market beta than a typical equity and more to the rate curve.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO: Steven O. Vondran — became President & CEO Feb 1, 2024, succeeding Tom Bartlett (who ran it through the rate storm). Internal promotion; ran the US Tower division prior. Comp: $1.0M base + 200% target bonus ($2.0M) plus equity.
• Track record: Vondran's tenure has been a "quality-of-earnings reset" — actively pruning low-return international (India, Australia, NZ, SA fiber, Mexico fiber, Poland), concentrating capital into developed markets + data centers, and running a cost program that has added >300bps of cash-EBITDA margin since 2022 (operating margin recovered 30% → ~46%). This is disciplined, shareholder-friendly capital allocation — the right playbook for a mature REIT.
• Capital allocation: Stated objective is to grow AFFO/share and ROIC over the long term. FY2025: $3.1B dividends ($6.72/share), $364.6M buybacks (2.04M shares), heavy debt repayment/refinancing, selective ~2,230 new sites + CoreSite development. Cumulative distributions to common ~$23.7B. Dividend just raised to $1.79/quarter ($7.16 annualized) — continued ~6%+ dividend growth.
• Skin in the game / insider ownership: n/a — precise insider % not sourced from on-disk filings (the DEF 14A is referenced but not in the research layer). Institutional ownership is high (typical large-cap REIT); founder-departed long ago — this is a professional-manager, not founder-led, company. Appropriate for the stage.
• Red flags: None systemic. Worth noting: the CoreSite acquisition (2021, $10.4B) was done at a cycle-peak data-center multiple and added ~$26.4B of goodwill+intangibles to the balance sheet (see Lens 10); and the Stonepeak ~29% / $2.5B preferred in the data-center business means minority leakage on the segment investors most want to own (NCI distributions reduce AFFO).

Lens 10 · Forensic Red Flags

Forensic lens — every figure filing-sourced where possible.

Balance-sheet / accounting risk points:

1. Goodwill + intangibles = $26.4B ($12.2B goodwill + $14.2B other intangibles) vs total equity of only $10.2B (and AMT common equity just $3.5B). The balance sheet is intangible-heavy from acquisitions (CoreSite, international). A data-center or international impairment would hit hard — and AMT already took $100.7M of impairments in FY2025 and $17.5M DISH impairment in Q1'26. Goodwill is a real watch-item, not a flag yet.
2. Negative/thin tangible equity + high leverage. Net debt ~$35B; net leverage ~4.9–5.1x net debt / annualized Adjusted EBITDA; covenant ceiling ≤6.0x (lien limit 3.5x). This is normal for a tower REIT but leaves limited cushion if EBITDA stalls and rates stay high. $6.1B current portion of long-term debt is a near-term refi at 4.7–5.35% coupons.
3. Straight-line revenue. AMT books contractual escalators on a straight-line basis, so reported revenue runs ahead of cash early in a lease and behind later. The $175.8M FY2025 / $34.9M Q1'26 straight-line declines in U.S. & Canada are why reported revenue looks worse than cash organic growth. AFFO strips straight-line out, so AFFO is the cleaner number — but GAAP revenue/EPS are flattered/depressed by this in ways that mislead a casual reader. This is the most important "accounting optics" item.
4. Non-controlling interests are large and growing. NCI equity $6.6B; ATC Europe is only 52%-owned (Allianz / La Caisse hold 48%), and Stonepeak owns ~29% of the data-center business. Consolidated revenue/EBITDA overstate AMT-attributable economics — always read AFFO attributable to common, not consolidated EBITDA. The Stonepeak mandatorily convertible preferred converts ~4 years from the Aug-2022 close (i.e., ~2026) — a dilution/NCI step-up to watch in the segment shareholders most prize.
5. Receivables build from disputes ($718.9M, up from $650.3M) — AT&T Mexico and DISH reserves. AMT took ~$30M (2025) + ~$10M (Q1'26) of AT&T Mexico reserves and expects more until the August 2026 arbitration.
6. SBC $58.4M Q1'26 ($174.2M FY2025) — modest (~2% of revenue), added back in Adjusted EBITDA but correctly subtracted in AFFO. Not a flatter-the-non-GAAP flag here.

Cash flow vs earnings: Cash from ops ($1,400.6M Q1'26) comfortably exceeds net income — normal for a D&A-heavy REIT. No divergence flag.

Regulatory findings (required):

• SEC Litigation Releases: None naming American Tower, 2021-06-23 → 2026-06-23 (EDGAR EFTS, LR).
• SEC AAERs: None in the search period (EDGAR EFTS, AAER).
• Non-SEC enforcement: No material FTC/DOJ/FCC consent decree, fine, or penalty surfaced for AMT in web search. The live legal matters are commercial disputes, not enforcement: (a) AT&T Mexico arbitration — AT&T Mexico (~$300M of 2025 tenant revenue) disputes the MLA lease calculation, withheld rents since early 2025; partial September-2025 settlement (escrow); hearing August 2026; AMT reserving until resolved; (b) DISH litigation — AMT filed in U.S. District Court (Colorado) Oct 2025 for declaratory judgment that DISH is not excused under the Strategic Collocation Agreement (DISH in default Jan 2026).
• Conclusion: No material regulatory or securities-fraud findings — verified via SEC EDGAR EFTS (LR, AAER), web search, and 10-Q Item 2/legal disclosures as of 2026-06-23. Risk here is counterparty/contract (DISH, AT&T Mexico), not regulatory.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (AFFO/share — the right metric for a REIT)

REIT-appropriate: project AFFO/share, not EPS. Built bottom-up from FY2025 actuals + FY2026 guidance. All `` with arithmetic; no forecast.ts logged (watchlist run).

Anchors: FY2025 AFFO (common) $5,041.6M, AFFO/share ~$10.78. FY2026 guided AFFO/share midpoint ~$10.99 (+$0.12 raise). Underlying organic tenant billings ~4% ex-DISH; DISH (~2% of property revenue) is a one-time 2026 drag; data centers +mid-teens.

Three-year path (FY2026E → FY2028E), AFFO/share:

Base-case logic: FY2026 is the trough growth year (~1% reported). Once DISH is lapped, AMT should return to mid-single-digit AFFO/share growth (4–5% organic tenant billings + ~3% escalators + data-center mid-teens + modest buybacks, net of higher refinancing cost). That is the "floor → reaccelerate" thesis management is selling. Brier forecast (not logged — watchlist): AMT FY2027 AFFO/share ≥ $11.50, p≈0.62, resolves 2028-02.

Valuation cross-check: at ~$184, P/AFFO ≈ 16.7x FY2026E. A re-rate to ~19–20x on FY2027E ~$11.65 → ~$220–235. Held flat at ~17x on FY2027E → ~$198. Downside if growth disappoints and the multiple compresses to ~14x on a $11.10 bear → ~$155.

Lens 12 · Bull vs Bear

Bull case. AMT is a toll road on global mobile data + a CoreSite AI call option, mispriced as a no-growth bond proxy. The moat is among the most durable in public markets (2% normalized churn, $54B contracted backlog, 84% US gross margins, irreplaceable footprint). Underlying organic is 4–5% — the ~1% reported is a transient DISH + straight-line + FX optical, not a structural break. Vondran's reset (prune international, concentrate on developed + data centers, +300bps margin, 6%+ dividend growth) is exactly right for the stage. CoreSite is the kicker: >1,000 land plots for multi-MW data centers, interconnection-dense, levered to AI inference — and arguably not in the ~17x multiple. Rates rolling over would re-rate the whole long-duration complex. A re-rate to historical ~20x AFFO + a 2027 reaccel = ~$220–235, +20–28%, plus a ~3.8% growing dividend.

Bear case (permanent-impairment candidates).

1. Structural deceleration, not an air-pocket. If US carrier capex stays muted (5G largely built), spectrum/efficiency reduces the need for new colocations, and LatAm consolidation churn persists past 2027, then ~1% AFFO growth is the new normal, not a floor — and a 17x multiple is then too high, not too low.
2. Rates stay higher for longer. $35B net debt, ~5x leverage, $6.1B near-term maturities refinancing at 4.7–5.35%. A bond-proxy whose interest cost keeps rising while growth stalls compresses both AFFO and the multiple — the 2022 playbook.
3. Carrier consolidation / bargaining power. Three US tenants are 85% of the biggest segment; a merger (or aggressive network-sharing) reduces the tenant count and lease pricing. DISH already proved a tenant can simply default.

Pre-mortem (18 months out, thesis broke): It's late 2027 and AFFO/share grew only ~1–2% again — DISH was lapped but LatAm churn and a soft US leasing cycle filled the hole; CoreSite growth slowed as utility power queues capped new capacity; rates didn't fall, so the refi drag bit and the multiple compressed to ~14x. The stock is ~$150. The error was believing 2026 was a floor when it was the second step down.

Are multiples too high? No — ~17x EV/EBITDA / ~16.7x AFFO is below AMT's history and a discount to CCI. The risk is earnings, not multiple: the multiple is fair if 2027 reaccelerates.

Contrarian view (what the market refuses to see): The market is anchored on the ~1% reported 2026 growth and treating AMT as a busted compounder. It is under-weighting (a) that ex-DISH organic is ~4–5% and DISH is a one-time clean-out, and (b) the CoreSite AI option. The mirror-image contrarian risk: bulls are anchored on "2027 reaccelerates" as if it's contractual — it is a forecast, and US tower leasing has structurally slowed.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the model: the US tower growth engine may simply be mature — 5G is largely deployed, and carriers extract more from existing spectrum (carrier aggregation, efficiency, eventually fixed-wireless/satellite offload) rather than hanging ever-more radios. If new colocations plateau, AMT becomes a ~CPI-plus-escalator annuity, not a grower — and shouldn't trade at 17x.
• Revenue concentration: 85% of the flagship segment is three carriers; 70% of Europe is one (Telefónica). DISH just demonstrated a large tenant can default and walk. A T-Mobile/Verizon network-sharing deal, or a Telefónica restructuring, would gut a segment.
• Why the moat is weaker than bulls think: switching costs protect the installed base, not growth. The moat guarantees the rent rolls; it does not guarantee new leasing. And the data-center "interconnection moat" competes with much-better-capitalized hyperscale and colo players, gated by utility power AMT doesn't control.
• Most dangerous competitor bulls underestimate: not CCI/SBA — it's the carriers themselves + spectrum efficiency + non-terrestrial (satellite direct-to-device) slowly eroding incremental tower demand, plus hyperscalers self-building data centers around CoreSite.
• Worst capital-allocation moves: CoreSite at $10.4B in 2021 at a peak data-center multiple, funded with debt right before rates spiked — and then selling 29% of it to Stonepeak (admitting it needed outside capital and diluting the best asset). The international empire-building of the 2010s that is now being unwound at a loss (India sold for $2.5B after billions invested) is a track record of buying high and exiting low.
• What must hold for today's price: ~4–5% organic reaccelerating in 2027, rates not rising further, no major tenant default/merger, CoreSite sustaining mid-teens despite power constraints.
• Growth disappoints 20–30%: if FY2027 AFFO/share comes in ~$11.10 (bear) vs ~$11.65 (base) and the multiple compresses to ~14x → ~$155, −16%, before dividend.
• Single scenario that permanently impairs: a structural step-down in US wireless capex (5G done, efficiency + satellite offload) coinciding with rates staying high — turning AMT into a slow-growth, over-levered bond proxy and forcing a multiple de-rate. Plausibility: moderate — the early signs (flat US organic ex-DISH is still only ~4.5%, decelerating) are real, but the contracted backlog and escalators put a hard floor under cash flows. Not a zero, but not a 2022-style 40% drawdown either.

Lens 14 · Management Questions (15, ordered by information value)

1. Ex-DISH and ex-straight-line, what is the true organic tenant-billings growth trajectory for U.S. & Canada over 2026–2028 — is ~4.5% the floor, and what specifically reaccelerates it?
2. Is 2026's ~1% reported AFFO growth a trough before reacceleration, or the start of structural deceleration? What's the bridge to your 2027 number?
3. How much incremental US colocation/amendment demand are you actually seeing from carriers now that 5G is largely deployed — units and dollars, not narrative?
4. On CoreSite: what is the realistic multi-MW capacity you can bring online through 2028 given utility power-queue constraints, and what's the signed backlog?
5. Why keep the Stonepeak 29% minority in the data-center business shareholders most want to own — will you buy it back, and what happens at the preferred conversion?
6. With $6.1B of near-term maturities refinancing at 4.7–5.35%, what is the blended interest-cost trajectory and its drag on AFFO/share through 2028?
7. What is your net-leverage target, and would you cut buybacks or slow the dividend to defend the credit rating if EBITDA stalled?
8. Latin America organic is guided ~−3% on consolidation churn through 2027 — when does it inflect, and is more of LatAm a divestiture candidate like India?
9. How exposed is incremental tower demand to fixed-wireless, satellite direct-to-device, and spectrum-efficiency offloading traffic away from macro towers?
10. Tenant concentration: what is your contingency if a major carrier merges or pursues aggressive network-sharing — and what did the DISH default teach you about credit underwriting?
11. AT&T Mexico arbitration (Aug 2026): downside scenario on the ~$300M revenue and the reserves if you lose?
12. What is the ROIC on CoreSite today vs the 2021 underwriting, and on new data-center development vs new towers?
13. How do you think about capital allocation between data-center development, buybacks, dividend growth, and debt reduction at the current ~17x AFFO and ~5x leverage?
14. What FX-hedging / repatriation strategy limits the quarter-to-quarter reported-revenue swings, and how much AFFO is structurally exposed to EM currencies?
15. What would make you exit another market or asset — and conversely, would you do another CoreSite-scale acquisition at today's cost of capital?

---