TMUS › analyze
TMUS · Analyze
Price at analysis: $177.75 | 52wk range: $165.66 – $258.66 | Mkt Cap $190.7B | Sector: Communication Services / Telecom Services
0. Knowledge check
Knowledge/INDEX.md and python .mcp/kb.py find TMUS returned no live note and no prior report — this
is a first look, researched fresh. Two Playbook entries were checked and are directly relevant and are cited
inline below: pitfall-yahoo-share-count-dual-class-fpi (checked — does not fire, see §1) and
pattern-margin-expansion-is-a-finite-growth-lever (fires — see §3). principle-down-a-lot-is-not-cheap was
also applied as the central valuation test (§3).
1. Fundamentals
Data-integrity check first (per pitfall-yahoo-share-count-dual-class-fpi): TMUS is single-class common
stock; Deutsche Telekom's ~50%+ stake is held in ordinary shares, not a super-voting class. marketCap ÷
price ($190.67B / $177.75 = 1.073B) reconciles with fin.py's SharesOut 1.07B — no dual-class /
share-count discrepancy. Insider% (56%) in the snapshot reflects Deutsche Telekom's control stake, not a
vendor error.
Second data flag found in this analysis: fin.py's snapshot OpMargin 0.25 does not reconcile
against the income statement (OpInc $18.56B / Revenue $88.31B = 21.0%, which matches roic.ai's
oper_margin of 21.01% for FY2025 exactly). The snapshot field is stale or built on a different
denominator — use 21.0% as the trustworthy operating margin, not the 0.25 snapshot field.
Third flag: fin.py's PE(fwd) 12.31 implies forward EPS of $14.44 — a 53% jump from ttm GAAP EPS
$9.46 in a single year. That is too large to be organic growth; it is almost certainly built on a
non-GAAP/adjusted EPS estimate for a later fiscal year than the label implies (consistent with
pitfall-vendor-forward-eps-is-the-wrong-fiscal-year). Forward P/E is not used in this valuation for
that reason — ttm P/E and EV/EBITDA (reconciled against roic.ai) are used instead.
Free cash flow — the real story
| FY | OCF | Capex | FCF | Buybacks | Dividends |
|---|---|---|---|---|---|
| 2022 | $16.78B | -$17.30B | -$0.52B | -$3.00B | $0 |
| 2023 | $18.56B | -$10.81B | $7.75B | -$13.07B | -$0.75B |
| 2024 | $22.29B | -$12.31B | $9.98B | -$11.23B | -$3.30B |
| 2025 | $27.95B | -$12.52B | $15.43B | -$9.97B | -$4.12B |
| 2026E (guided) | — | — | $18.4-18.8B | — | — |
FCF went from negative to nearly $15.4B in three years as Sprint-merger integration capex rolled off (capex fell from $17.3B in 2022 to ~$12.5B, now stabilizing). A clean 5-8yr CAGR isn't computable — 2022's base was negative — so treat 2023→2025's ~41%/yr FCF growth as merger-normalization, not a repeatable rate. The 2026 guide ($18.4-18.8B, +19-22% y/y) already shows the expected deceleration, which is still a healthy trajectory.
FCF/share: $13.64 (2025) → implied P/FCF ≈ 13.0x at $177.75, materially below roic.ai's own FY2025 low-of-year multiple of 14.2x and FY2024's average of 22.1x. This is a genuine below-own-band signal (see §3), not just a drawdown from the 52wk high.
Capital allocation (FY2025, $27.95B OCF)
Growth capex -$12.52B · Buybacks -$9.97B · Dividends -$4.12B · Acquisitions -$7.58B (fiber JV stakes: GoNetspeed/Greenlight, i3 Broadband) · Debt issued $12.01B vs repaid -$7.45B (net debt +$4.56B, funding the fiber build-out). ~91% of FCF ($14.09B of $15.43B) returned to shareholders via buybacks + dividends, with the fiber expansion funded by incremental debt rather than crowding out shareholder return — a reasonable allocation given still-low leverage headroom (see below), though it means net debt is not shrinking while the fiber bet plays out.
Leverage
Debt/Assets 55.8% · Net Debt/EBITDA 2.54x · Gross Debt/EBITDA 3.85x · EBITDA/Interest 8.4x (roic.ai, FY2025, cross-validated). Investment-grade profile for the industry; current ratio 0.92 is normal for a telecom (handset financing receivables, no meaningful working-capital risk).
Share count & growth
Diluted shares fell 1.26B (2022) → 1.13B (2025), a -3.4%/yr 3yr CAGR — real buyback-driven shrinkage, a genuine per-share tailwind. Revenue CAGR 3.5-3.6%/yr (3yr, total revenue incl. equipment); service revenue is guided faster (~8-9%/yr, $77.0B 2026E → $80.5-81.5B 2027E). Net income's 61.9% 3yr CAGR is an artifact of 2022's abnormally low Sprint-integration-cost base ($2.59B) — exclude it from any forward extrapolation. Dividend: initiated late 2023, raised every year since — $0.65 → $0.88 (+35%) → $1.02 (+16%) quarterly — too young for the 8-10yr CAGR the framework normally wants; note as data-limited, not absent.
Scorecard: Strong and accelerating FCF, real per-share buyback tailwind, investment-grade leverage, young but fast-growing dividend with ample coverage room (FCF payout ~27%). The soft spot is total revenue growth (3.5%) trailing the service-revenue growth the bull case leans on, and a leverage build now funding fiber optionality rather than debt paydown.
2. Moat & Competitive Advantage
ROIC: 7.8% (roic.ai return_on_inv_capital), stable 2024→2025 (7.79%→7.82%). Modest in absolute
terms — telecom is capital-intensive — but positive and stable against a WACC likely in the 6-7% range for
an investment-grade, low-beta (0.33) utility-like issuer. Gross margin essentially flat (63.6%→62.9%),
a small compression worth watching given the price war described below, not yet a moat-erosion signal.
Moat sources, named explicitly: - Switching costs — contracts, family/multi-line plans, device financing, autopay bundling, number porting friction. Real but not insurmountable — the entire industry lives on this. - Cost/scale advantage — the 2.5GHz Sprint spectrum gives T-Mobile a structural mid-band capacity edge over Verizon and AT&T. This is the actual, measurable source of its network lead: Ookla rated T-Mobile the #1 mobile network and #1 5G network for 1H 2026 (median 5G download 314 Mbps vs. Verizon 214 Mbps, AT&T 173 Mbps). - Regulatory scarcity — spectrum licenses are government-granted, quasi-fixed-supply assets. This is why SpaceX has to spend $19.6B on its own spectrum and years of satellite buildout just to attempt entry — the barrier is real, not just competitive inertia. - No classical network effect — interconnection is mandated industry-wide; this is not a winner-take-most dynamic the way software network effects work.
Adversarial stress-test — "how would a well-funded rival attack?" Two attacks are live right now, one is live but distant: 1. Cable MVNOs undercutting on price (Comcast Xfinity Mobile, Charter Spectrum Mobile, both riding Verizon's wholesale network) plus AT&T/Verizon re-entering the entry-level tier T-Mobile used to own alone (Wolfe Research's downgrade thesis: T-Mobile's Essentials Saver 2.0 at $50/1-line is now priced above AT&T's $15 entry point and Verizon's $30/line). This is a genuine erosion of the "value leader" half of T-Mobile's historical differentiation — the network-quality half is intact and still widening, the price half is not. 2. SpaceX Starlink Mobile direct-to-cell — real, well-funded, explicitly stated ambition to "win customers from the Big Three." But T-Mobile's own CFO argues (and independent analysts largely concur) that satellite capacity per cell makes it a rural/gap-filler, not a dense-urban competitor, at least until Starlink Mobile V2 launches (2027) and full service (end of 2027) — this is a 2027+ risk, not a 2026 one, but it is the one genuine long-horizon disruption vector worth tracking. 3. Capacity allocation tension — T-Mobile is simultaneously growing FWA (5G Home Internet, 8.5M connections, +31% y/y, targeting 12M by 2028 / 15M by 2030) and mobile data demand on the same spectrum. FWA is a genuine growth engine and a hedge against cable's broadband bundle, but it competes with mobile customers for the same finite mid-band capacity that is the core network-quality moat.
Disruption forecast (5-10yr arc): the near-term disruption is competitive-intensity (price), not technological — the network-quality edge from Sprint spectrum is durable through this decade. The longer-dated technological disruption (satellite direct-to-cell at scale) is real optionality risk but not an imminent threat per either management or the balance of analyst opinion.
Evergreen assessment: connectivity is a genuine forever-need, and the US wireless market is a durable 3-scaled-player oligopoly with high regulatory and capital barriers to true new entry. Call it evergreen with a live, real competitive-intensity risk factor — not immune to margin/share trades within the oligopoly, which is exactly what is being priced right now. Not a 9-10/10 fortress; a solid 7/10 durable infrastructure business.
3. Valuation
Applying the framework conditionally: TMUS is a mature, dividend-paying, moderate-growth infrastructure name — Graham gets little weight (reason below), Bogle and the below-band multiple test carry the load, DYT/DDM apply but are data-limited by a dividend history under 3 years old.
Graham's Intrinsic Value — low weight, explained
√(22.5 × 9.46 × 52.35) = $105.56 vs. price $177.75 — Graham says ~68% overpriced. This number is not
trustworthy here and is given near-zero weight: Graham's book-value term assumes clean, unlevered
tangible assets, but TMUS's GAAP book value is dominated by Sprint-merger goodwill/intangibles, and
roic.ai shows tangible book value per share is negative (pr_to_tang_bv_per_sh: -3.99). A formula
built for asset-heavy industrials with a clean balance sheet is the wrong tool for a spectrum-and-goodwill
balance sheet funded by years of aggressive buybacks that have shrunk book equity. Cited as required
context, not as a load-bearing signal.
The load-bearing test: is TMUS actually cheap against its own history? (per principle-down-a-lot-is-not-cheap)
| Multiple | Current | FY2025 own-year range (roic.ai) | FY2024 own-year range |
|---|---|---|---|
| EV/EBITDA | 8.97x | 10.4x – 13.3x (avg 11.9x) | 9.4x – 12.7x |
| P/FCF | ~13.0x | 14.2x – 20.3x (avg 17.3x) | 18.7x – 29.2x |
| P/E (ttm) | 18.79x | 19.9x – 28.4x (avg 24.2x) | 16.4x – 25.6x |
TMUS is trading below every one of its own trailing-8-quarter multiple floors, not just off its 52wk high. This is the genuine pass the principle asks for — down 31% from the Feb 2026 high and below its own cheapest 2024-2025 print on three independent multiples, with a growing, not shrinking, earnings base (revenue, FCF and EBITDA are all still guided up in 2026-2027). That combination is exactly the "down a lot and below its own band and the denominator hasn't reset" case the principle calls a genuine signal, not a value trap.
pattern-margin-expansion-is-a-finite-growth-lever — applied to the bull case
Management's Feb 2026 Capital Markets Day raised multi-year guidance, and some sell-side models show EPS growing ~65% over 3 years (~18%/yr) against ~4-5%/yr underlying revenue growth. That gap is bridged by margin expansion (2026E Core Adjusted EBITDA guided $37.0-37.5B vs FY2025's $31.79B, +17% in one year) and continued buybacks — both finite levers, per the pattern. EBITDA margin cannot expand at that clip indefinitely; once Sprint-synergy-driven margin gains and the buyback pace both normalize, EPS growth converges toward the ~4-5% underlying revenue growth. Anchor any terminal multiple on that ~4-5% figure, not the double-digit EPS algorithm.
Bogle's Expected Return
Dividend yield (2.3%) + realistic underlying earnings growth (~8-10%/yr — revenue growth plus a fading but not-yet-exhausted margin/buyback tailwind, well short of the 18%/yr headline) ± P/E normalization (ttm P/E at 18.8x sits below its own 2024-2025 average of ~19.4-24.2x, so multiple normalization is a plausible tailwind, not a headwind, if the current de-rating proves sentiment-driven). Expected return ≈ 10-15%/yr — attractive for a low-beta (0.33) name, conditional on the margin/buyback deceleration being gradual rather than abrupt.
DYT / DDM — applicable, but data-limited
Dividend history is under 3 years old — too short for the framework's normal multi-year yield band, so DYT
is supportive context only. Checked for the DYT-inversion trap explicitly (pitfall-dyt-inverts-when-
price-caused-the-yield): the current yield rose from both a falling price (-31% from high) and a
genuinely growing dividend (+57% cumulative since initiation) — the good-signal combination, not the
inverted one.
Two-stage DDM: $4.08 current annualized dividend, Stage 1 (yrs 1-5) at 12%/yr growth (payout ratio still only ~27% of FCF, ample room), terminal growth 4% (matching underlying revenue growth), discount rate 7.5-8%:
| Discount rate | DDM value |
|---|---|
| 8.0% | ~$150 |
| 7.5% | ~$172 |
DDM alone (dividends only) undercounts total return — it ignores the buyback, which is currently a larger capital-return channel than the dividend ($9.97B vs $4.12B in 2025). Treat the DDM range as a floor-ish anchor, not the full picture.
Fair value synthesis
Weighing the below-own-band multiple test (strongest signal), Bogle's expected return, and DDM as a floor check, against Graham's near-zero weight: fair value ≈ $185-230. Current price ($177.75) sits at or just below the low end — consistent with genuine, not superficial, "quality on sale." Sell-side consensus mean target ($243.38) sits above this range; targets were broadly cut in July 2026 (Morgan Stanley $260→$230, Barclays $230→$215, KeyBanc $260→$250) but remain mostly well above spot even after the cuts — Wells Fargo's $169 is the lone target below current price.
Entry: $165-185 (the current price and 52wk-low zone already qualify). Trim: 24x ttm P/E — roughly TMUS's own FY2025 average multiple, using ttm (not forward) P/E specifically because the vendor's forward EPS field is not trustworthy here (see §1 flag). At $9.46 ttm EPS that is ~$227 today, and the multiple convention means the dollar level rises with EPS as reported in each subsequent quarter.
4. Sentiment — why is it near its 52wk low right now?
This is a 9-month grinding de-rating, not a single-week panic. TMUS peaked ~$270 in Feb-Mar 2026, then declined steadily through the summer to a $165.66 low in June 2026, currently ~$177.75 — a -34% move over roughly six months, with volume rising on down-months (June 2026: 132M shares vs. a ~85-100M typical month). That shape matters: it reads as a sustained sentiment/positioning shift, not a single shock.
Three concrete, dateable drivers, in order of how much they explain:
- Rate-plan-modernization backlash (the proximate, largest driver). On 2026-06-29, T-Mobile forced ~4 million legacy 3G/4G-era plan customers onto new pricing (+$4-6/line), eliminating the KickBack discount; some customers held written "we won't raise your rate, ever" documentation. On the 2026-07-23 Q2 call, CFO Peter Osvaldik guided Q3 postpaid net adds down to just ~250,000 (from 277,000 in Q2 and a 950k-1.05M full-year target), explicitly flagging temporarily elevated churn. Q3 (reporting 2026-10-22) is the print that will show whether this was a manageable, priced-in blip or a deeper erosion of the value-brand promise.
- Wolfe Research downgrade (2026-08-14, Outperform → Peer Perform). The rationale is the moat's live soft spot from §2: AT&T/Verizon have re-entered the entry-level price tier T-Mobile used to own alone, "long-term revenue growth forecast risk tilts negatively as competition expands in T-Mo's core," and broadband/6G capex could pressure capital returns and leverage. Every other major sell-side desk (UBS, TD Cowen, Barclays, KeyBanc, RBC, Morgan Stanley, Scotiabank) cut price targets through July 2026 on the same competitive-intensity read, while keeping Buy/Overweight ratings — a genuine, broad-based trim-not-flee signal.
- SpaceX/Starlink Mobile headline risk — the most-discussed but least near-term-relevant driver. Recent headlines (SpaceX's $19.6B spectrum spend, "expects to win customers from the Big Three") read as scary but management and independent analysts (Bernstein, BofA) largely agree the physical capacity constraints make this a 2027+ complementary-coverage story, not an imminent competitive substitute. This is overweighted in the headline flow relative to its near-term earnings impact.
Verdict on sentiment: predominantly cyclical/competitive-intensity repricing, with one real structural watch-item. The rate-plan backlash and the entry-tier price war are genuine, current, and could legitimately erode T-Mobile's "best value + best network" dual differentiation if AT&T/Verizon sustain the undercutting — that is a real risk, not noise. But it is happening while revenue, FCF and EBITDA guidance are all being raised, not cut, and the network-quality half of the moat (Ookla #1 in 1H 2026) is unaffected. That combination — sentiment falling while forward guidance rises — is the core of the "quality on sale" read, tempered by a genuine, dateable near-term risk (Q3 churn) rather than a manufactured one.
5. Synthesis — weighted verdict
Company type: mature, dividend-paying, moderate-growth infrastructure/oligopoly name. Per the framework's conflict-resolution rule, Fundamentals + Valuation carry the most weight here; Moat and Sentiment are used to explain the why now and to bound the downside case, not to override the valuation read.
What agrees: Fundamentals (accelerating FCF, shrinking share count, investment-grade balance sheet, fast-growing young dividend) and Valuation (the stock is below its own 2024-2025 multiple floor on three independent measures, with a growing not shrinking earnings base) point the same direction — genuine, not superficial, value. Sentiment confirms the drivers are dateable and largely priced-competitive rather than an unexplained air-pocket.
Named tension — where Moat and Sentiment push back against Fundamentals/Valuation: the bull case's implied EPS algorithm (mid-teens+ growth) leans heavily on margin expansion and buybacks, both finite levers per the Playbook pattern applied in §3 — the underlying business is a ~4-5%/yr revenue grower, and the entry-tier price war (Wolfe's thesis) is a live, current erosion of one half of the historical differentiation. This isn't a reason to avoid the name — it's a reason the fair-value range is $185-230 and not $250+, and a reason to size the position for a mature-compounder return profile (high single to low double digits, dividend-assisted), not a re-rating story.
Genuine risk factors, named: - Q3 2026 churn (print 2026-10-22) could show the rate-plan backlash was worse than management's own guide — the single nearest-dated falsifier of this thesis. - Sustained AT&T/Verizon/cable-MVNO price competition at the entry tier could structurally compress the "best value" half of the moat even if network quality stays #1. - Leverage is being used to fund fiber JV growth rather than delever — fine at 2.5x net debt/EBITDA today, but worth re-checking if rates or competitive spend pressure the balance sheet further. - SpaceX/Starlink Mobile is a real, long-dated (2027+) optionality risk, not yet an earnings risk.
Verdict: ACCUMULATE, conviction 6.5/10. Good company, and the market has priced in more fear than the guided fundamentals currently justify — but the fear (entry-tier price war, Q3 churn) is real enough that this is a scale-in accumulate at $165-185, not a full-conviction back-up-the-truck buy. Re-check hard at the 2026-10-22 Q3 print.
Prepared by Financebot — Fundamentals, Moat, Valuation, and Sentiment analyst lenses, synthesized.