AI-generated analysis, not investment advice. The articles are written by AI, edited, and checked against company filings — but the judgements are opinions and the figures go stale. How this is made · Terms
The MoatWide moat
T-Mobile US (TMUS) — moat facet
Spectrum keeps US wireless to three carriers and T-Mobile wins inside that market, but the licences hold its return on capital at 7.6%, just under an 8% hurdle.
T-Mobile's moat has two layers. The outer one is shared: spectrum licences that the government rarely issues, which keep American wireless to three national carriers. The inner one is T-Mobile's own: a network now ranked first on several surveys, sold at a lower price than Verizon's, run at a scale that makes about half of service revenue into EBITDA.
Four facets carry the argument. Spectrum is the asset nobody can copy, $98,032 million of licences at the end of 20251. The Network and the Price is how T-Mobile wins customers inside the three-carrier market. Scale and the Cost Base turns the customers into cash. The Account, Not the Phone is why the customers stay.
The return on capital shows the moat and its cost together. Measured the repo way, return on invested capital was 4.1% to 6.6% from 2015 to 2019 except for 9.0% in 2017, when tax reform flattered it, fell to 3.0% in 2022 after Sprint and its merger costs, and recovered to 7.6% in 2024 and 20252. That is just below an 8% hurdle.
The reason is the licences. Spectrum is never amortised, so it stays in the denominator at cost; it was 51.8% of invested capital in 20253. Excluding it, the return was about 16.3%4. The network earns well on what it uses; the company earns roughly its cost of capital on what it paid for the right to use it.
The moat is rated wide because the three-carrier structure is protected by law and scarcity, and T-Mobile is the strongest of the three on growth. The trajectory is stable: the network and price advantages widened through 2025, while the slower account growth, rising churn and the end of phone reporting in 2026 make the next step harder to read.
Third parties measure the same return more generously. stockanalysis.com puts T-Mobile's trailing return on invested capital at 8.93% and its weighted average cost of capital at 4.60%5, which would put the company comfortably above its hurdle. The difference is method: a lower assumed cost of capital and a different definition of invested capital. On Moat Explorer's method and an 8% hurdle, T-Mobile is close to the line rather than well over it.
The falsifier is the return on capital. If it stayed below 8% after the UScellular synergies are complete at the end of 2027, the wide moat would be protecting a business that does not earn its cost of capital.
ROIC 3.0% (2022) to 7.6% (2024-2025); churn and organic growth softening.
The whole moat in one figure; staying below 8% after the UScellular synergies would mean the moat does not earn its cost of capital.
- ReportedSpectrum is the asset nobody can copy, $98,032 million of licences at the end of 2025.T-Mobile US, Inc. Form 10-K for fiscal 2025 (year ended 31 December 2025) - spectrum: holdings by band, licence transactions and the spectrum roll-forward. — FY2025 · publ. 11 February 2026 · source ↗
- Moat Explorer calcMeasured the repo way, return on invested capital was 4.1% to 6.6% from 2015 to 2019 except for 9.0% in 2017, when tax reform flattered it, fell to 3.0% in 2022 after Sprint and its merger costs, and recovered to 7.6% in 2024 and 2025.Moat Explorer calculation, repo method (tools_roic_edgar.py logic run on CIK 1283699): return on invested capital 4.1% (2015), 5.3%, 9.0%, 6.6%, 6.4% (2019), 4.3%, 3.6%, 3.0%, 5.9% (2023), 7.6%, 7.6% (2025). 2025: operating income 18,279 x (1 - 3,289 / 14,281 = 23.0% tax) = NOPAT 14,069; invested capital (total assets less current liabilities less cash) 182,452 (2024) and 189,139 (2025), average 185,796; 14,069 / 185,796 = 7.6%. Spectrum licences 100,558 (2024) and 98,032 (2025); invested capital excluding them 81,894 and 91,107, average 86,501; 14,069 / 86,501 = 16.3%. Spectrum licences over 2025 invested capital 98,032 / 189,139 = 51.8%. 2017 is flattered by the tax-reform benefit; 2020-2022 are depressed by the Sprint spectrum step-up and merger costs. — FY2015-FY2025 · publ. October 2026 · source ↗Method: NOPAT (operating income times one minus the effective tax rate, clamped 0-35%) divided by average operating invested capital (total assets less current liabilities less cash and equivalents); SEC EDGAR XBRL. Spectrum licences are indefinite-lived and are not amortised, so they stay in the denominator at cost.
- Moat Explorer calcSpectrum is never amortised, so it stays in the denominator at cost; it was 51.8% of invested capital in 2025.Moat Explorer calculation, repo method (tools_roic_edgar.py logic run on CIK 1283699): return on invested capital 4.1% (2015), 5.3%, 9.0%, 6.6%, 6.4% (2019), 4.3%, 3.6%, 3.0%, 5.9% (2023), 7.6%, 7.6% (2025). 2025: operating income 18,279 x (1 - 3,289 / 14,281 = 23.0% tax) = NOPAT 14,069; invested capital (total assets less current liabilities less cash) 182,452 (2024) and 189,139 (2025), average 185,796; 14,069 / 185,796 = 7.6%. Spectrum licences 100,558 (2024) and 98,032 (2025); invested capital excluding them 81,894 and 91,107, average 86,501; 14,069 / 86,501 = 16.3%. Spectrum licences over 2025 invested capital 98,032 / 189,139 = 51.8%. 2017 is flattered by the tax-reform benefit; 2020-2022 are depressed by the Sprint spectrum step-up and merger costs. — FY2015-FY2025 · publ. October 2026 · source ↗Method: NOPAT (operating income times one minus the effective tax rate, clamped 0-35%) divided by average operating invested capital (total assets less current liabilities less cash and equivalents); SEC EDGAR XBRL. Spectrum licences are indefinite-lived and are not amortised, so they stay in the denominator at cost.
- Moat Explorer calcExcluding it, the return was about 16.3%.Moat Explorer calculation, repo method (tools_roic_edgar.py logic run on CIK 1283699): return on invested capital 4.1% (2015), 5.3%, 9.0%, 6.6%, 6.4% (2019), 4.3%, 3.6%, 3.0%, 5.9% (2023), 7.6%, 7.6% (2025). 2025: operating income 18,279 x (1 - 3,289 / 14,281 = 23.0% tax) = NOPAT 14,069; invested capital (total assets less current liabilities less cash) 182,452 (2024) and 189,139 (2025), average 185,796; 14,069 / 185,796 = 7.6%. Spectrum licences 100,558 (2024) and 98,032 (2025); invested capital excluding them 81,894 and 91,107, average 86,501; 14,069 / 86,501 = 16.3%. Spectrum licences over 2025 invested capital 98,032 / 189,139 = 51.8%. 2017 is flattered by the tax-reform benefit; 2020-2022 are depressed by the Sprint spectrum step-up and merger costs. — FY2015-FY2025 · publ. October 2026 · source ↗Method: NOPAT (operating income times one minus the effective tax rate, clamped 0-35%) divided by average operating invested capital (total assets less current liabilities less cash and equivalents); SEC EDGAR XBRL. Spectrum licences are indefinite-lived and are not amortised, so they stay in the denominator at cost.
- Third-party estimatestockanalysis.com puts T-Mobile's trailing return on invested capital at 8.93% and its weighted average cost of capital at 4.60%, which would put the company comfortably above its hurdle.stockanalysis.com, T-Mobile US statistics, October 2026: 1.07 billion shares, P/S 1.92, P/B 3.15, enterprise value $294.21bn, EV/EBITDA 8.56, trailing free cash flow $18.40bn, price down 28.50% in 52 weeks, shares outstanding down 3.93% in a year, return on invested capital 8.93% and a weighted average cost of capital of 4.60% (stockanalysis estimates). — October 2026 · publ. 5 October 2026 · source ↗