The call
BUY AT&T (T) $21–23, Target $28 (stretch $30), Invalidation weekly close <$19.50, 1–3 months / 6–12 months, SPECULATIVE-HIGH conviction on income + event. Live $22.88 (+1.7% Oct 13).
Why now
Bernstein cut T to Market Perform, PT $24, on SpaceX's $8B 800 MHz deal — the last Bernstein outperform in telecom gone. Fear, not fundamentals, knocked T to 7× forward earnings.
The disagreement
Consensus sees a fourth national carrier in orbit. I see a $67B+ 10-year terrestrial build, FCC approvals, and physics that still loses to fiber + towers in cities. Starlink solves the last 2%, as Stankey said.
The level that changes everything
$19.50 weekly close kills the thesis (52-week low $19.89 breaks). Above $24 into Oct 21 Q3 (EPS est. $0.61) confirms fiber re-acceleration.
First sentence: On Oct 8–9 SpaceX agreed to buy Grain Management's nationwide 800 MHz portfolio (~14 MHz paired) for ~$8B, plus FCC approval for 15,000 Gen2 satellites — and telecom fell 6–7% in a day.
Details matter. Forbes (Oct 9) reports AT&T -6.6% to $23, Verizon -7.2% to $43, T-Mobile -6.8% to $159.70 premarket Friday. That wiped ~$39–45B off telecom caps in two sessions. Then Bernstein's Laurent Yoon on Oct 12–13 downgraded AT&T to Market Perform, PT $24 from $25, saying "ongoing — and justifiable — pressure from SpaceX's foray into both broadband and wireless" plus Q4 seasonality makes an Outperform "increasingly difficult... as volatility seems inevitable."
This was the last domino: Yoon no longer recommends any Big Three carrier. Morningstar raised uncertainty to High. Piper cut AST SpaceMobile 45% on the same fear. The S&P Telecom group logged its second-worst day ever Friday.
My opinion: this is classic overhang repricing, not earnings impairment. Bernstein is right that debate will be "prolonged" — which is exactly why it creates entry. The stock is down >9% YTD to $22.88 while the business beat in Q2 and reiterated $18B+ FCF guidance.
First sentence: Starlink Mobile is a rural coverage product trying to become an urban capacity business — that inversion costs tens of billions and years.
Three load-bearing facts, all from primary sources:
Physics: Low-band 800 MHz travels far and penetrates buildings — fixing "one of the key remaining technical gaps," per SpaceX — but capacity per square mile in Manhattan or Dallas still belongs to dense fiber-fed small cells. Verizon CEO Dan Schulman and AT&T CEO John Stankey both made the same point this month: satellite "solving the last 2%, not... the 98%." T-Mobile's CEO called the threat overstated. Even SpaceX-friendly analysts concede it needs terrestrial leases or a build.
Capex: Morgan Stanley models ~$200B global connectivity capex in 5 years, 80% on mobile incl. spectrum; Wells Fargo estimates $67B over 10 years for a U.S. land network alone. SpaceX paid $17B to EchoStar for 2 GHz last year plus $8B now — before a single tower lease. AT&T already spends $23–24B per year and owns 38.6M fiber locations.
Time + regulation: Grain deal needs FCC approval; EchoStar sale carried unusually restrictive conditions; upper C-band auction ahead is the tell. Citi's Heath Terry calls a full build "more likely" given Musk's preference — and a net-positive for towers (AMT/CCI/SBAC), not an instant share loss. Pivotal's 13% share by 2036 bull case is a decade away — you collect ~$3.33 in dividends before then at today's yield.
First sentence: Q2 2026 (July 22 primary source) proves the pivot — Advanced Connectivity is growing 5%+ while legacy copper is being shut off.
Read the Q2 release, MD&A logic: revenue $31.6B +2.3%, adjusted EPS $0.65 vs $0.54 (+20%), adjusted EBITDA $12.3B +5.2%, FCF $4.7B vs $4.4B. Advanced Connectivity service revenue $23.5B +5.1%, operating income +20.3% to $7.3B. Net adds: 367k fiber + 279k fixed wireless = 646k broadband, plus 432k postpaid phone at 0.86% churn — best consumer postpaid account growth in 3+ years.
Scale: 38.6M fiber locations, on track >40M by end-2026 and >60M by 2030 (Lumen mass-market fiber closed Feb 2 + Forged Fiber 37 / Gigapower JV Oct 6 + $3B Corning deal). Convergence — 42.5% of advanced home internet households also hold AT&T wireless — is the moat satellite can't replicate: one bill, lower churn, higher lifetime value.
Capital returns do the waiting math: $1.11 dividend (6.85% yield at $22.49 close Oct 12), ~$10B buybacks in 2026 of $24B through 2028, $45B+ total returns 2026–28, FCF $18B+ → $19B+ → $21B+. Net debt $126.4B, targeting 2.5× leverage post-EchoStar close.
| Metric (live Oct 13) | T (AT&T) | VZ (Verizon) | TMUS |
|---|---|---|---|
| Price | $22.88 | $43.23 | $153.62 (52-wk low $147.90) |
| P/E TTM / Forward | 7.1× / 7.0× | 10.9× / 10.3× | 15.5× / 14.9× |
| Dividend yield | 6.85% | 7.11% | 1.75% |
| Beta / 52-wk range | 0.26 / $19.89–$29.44 | 0.26 / $38.39–$51.68 | 0.38 / $147.90–$231.02 |
| Growth signal | Fiber + Lumen, 5%+ svc rev | Frontier fiber, 4% svc guide | 9.7% rev growth, pricey |
My opinion: At 7× forward with double-digit EPS CAGR guide ($2.25–2.35 in 2026), T is priced for perpetual decline while delivering mid-single-digit converged growth. Verizon screens similarly cheap but carries Frontier integration + higher leverage; T-Mobile grows faster but at 15× with Starlink partnership complexity. For a diversifier beyond tech, T is the income anchor.
First sentence: This is an event-income trade — 6.8% yield + $10B buyback floor + Oct 21 Q3 ($0.61 EPS est.) vs. a fear overhang with no near-term revenue impact.
| Field | Level |
|---|---|
| Entry | $21–23 (add on $23 Friday flush retest; live $22.88) |
| Target | $28 base (22% upside + yield = ~29% total), stretch $30 into 2027 buyback acceleration |
| Invalidation | Weekly close <$19.50 (below 52-wk low $19.89 + dividend-cut signal) |
| Timeframe | 1–3 months to Q3/Q4 print + FCC headline; 6–12 months to $28 |
| Conviction / Size | SPECULATIVE-HIGH (income + event, single signal: overhang mispricing) — 2–3% position, half now half on <$22 |
| Audience | Income + diversifier sleeve; pairs with EQ-TOWER-1 long |
Bull (30%): Q3 fiber >350k, Lumen penetration lifts toward 40%, FCF confirms $18B+ → multiple to 9× 2027 EPS (~$2.60) = $28–30 by Q1. Base (50%): in-line 300k fiber, $0.61–0.63 EPS, FCC delays Grain → drift to $25–26 (Bernstein PT + yield). Bear (20%): postpaid phone <200k + broadband ARPU cut on Starlink pricing chatter → $19.50 test, dividend safety questioned.
What to do: Nibble T $21–23, sell VZ covered calls if overweight telecom, hold AMT/CCI as SpaceX-build hedge. No chase >$24 pre-earnings (Bernstein PT caps near-term). Add aggressively only if Oct 21 confirms converged churn <0.90%.
First sentence: If AT&T cuts the $1.11 dividend or loses postpaid subs two quarters straight, the fiber story is broken — exit.
Falsification: (1) Weekly close <$19.50; (2) Q3 Oct 21 or Q4 fiber net adds <200k with management cutting 2026 FCF <$17B; (3) FCC fast-approves Grain + SpaceX signs a wholesale deal with Verizon at deep discount, collapsing wireless ARPU >3% YoY. Any one triggers review; two = exit.
Risks quantified: $144B total debt — every 100bp on refinancing ≈ $1.4B interest drag vs $18B FCF (manageable unless 10Y >5.5%); copper decommissioning past 2029 delays strand costs (legacy EBITDA turns negative after 2027 by guide); Starlink 7M subs by 2027 (Morgan Stanley) could shave ~100bp off broadband pricing — offset by convergence if >45%.
Accountability: Our EQ-MU-1 long ($980 daily kill) stopped today — MU $938.82 -5.1% on Korea CXMT rout (KOSPI -5–7%, SK Hynix/Samsung rout spilling to MU/STX/WDC). That's a loss, thesis broken near-term on China memory oversupply fear. I keep the 6–12M $1,536 math but stand aside per discipline — no averaging a broken level. Full memory-rout review tomorrow with CPI Oct 14.