T-Mobile US TMUS

Reference table

1. What the business is, and how the disclosure actually works

T-Mobile US, Inc. is headquartered at 12920 SE 38th Street, Bellevue, Washington. The profile lists Srinivasan Gopalan as chief executive — corroborated by the 8-K of 2026-07-07, which refers to "Srinivasan Gopalan, the Company's President and Chief Executive Officer." The 10-K states the company employed approximately 75,000 full-time and part-time employees at 2025-12-31, matching the vendor's fullTimeEmployees field exactly. Listed on NASDAQ Global Select; CIK 0001283699; CUSIP 872590104; ISIN US8725901040; common stock par value $0.00001.

A note on the vendor's company description, which is badly stale. It states network scale "As of December 31, 2021" and a subscriber count of approximately 108.7 million, and it does not mention UScellular, the fibre joint ventures, or the change of chief executive. It should not be used for any factual claim in this dive and it is not.

One reportable segment — which settles the seg_prod question before it arises

The 10-K states, in Note 1: "We operate as a single operating segment." The June 10-Q's Note 11 – Segment Reporting says "The following table provides the operating financial results of our Wireless segment" — singular.

This matters because it makes the usual segment/product confusion impossible in one direction and certain in the other. The vendor's seg_prod block cannot be the reportable-segment table, because the reportable-segment table has one line. What seg_prod actually is, is the revenue-disaggregation table from Note 11 – Revenue from Contracts with Customers. We say so explicitly rather than presenting it as a segment table.

And it reconciles exactly. We checked every year.

Four consecutive years summing to the dollar. And every line change matches the 10-K MD&A verbatim: "Postpaid revenues increased $5.6 billion, or 11%" (we compute +$5.592B, +10.7%); "Prepaid revenues increased $98 million, or 1%" (+$98M, +0.9%); "Wholesale and other service revenues decreased $562 million, or 16%" (−$562M, −16.3%); "Equipment revenues increased $1.7 billion, or 12%" (+$1.709B, +12.0%); "Other revenues were essentially flat" (+$72M). This is one of the cleanest vendor-to-filing reconciliations in this batch and it is worth recording as such.

One exception, and we flag it rather than smooth it. The FY2021 row does not sum: five lines total $77.795B against reported FY2021 revenue of $80.118B, a $2.323B shortfall (2.9%). The FY2020 and FY2019 rows contain a sixth line, "Roaming and Other Service Revenue" ($2.078B in FY2020), which is absent from FY2021 onward. The FY2021 disaggregation row is incomplete and is not used anywhere in this dive.

seg_geo is EMPTY. No geographic breakdown is supplied. T-Mobile operates in the United States, Puerto Rico and the US Virgin Islands only, so the economic loss is small — but the field is blank and we state that rather than implying we checked it.

The three eight-Ks — read in full, and none of them is what the pattern predicted

The archive contains three 8-Ks filed inside five weeks. A cluster like that on a company with $86.9B of debt normally signals a financing or an acquisition. It does not here, and we read all three end to end to establish that.

• 8-K filed 2026-06-18 (event 2026-06-16): the annual meeting of stockholders. Thirteen directors elected, Deloitte & Touche ratified as auditor for fiscal 2026, say-on-pay approved. The vote tallies are in tables the extraction stripped; the file contains no numbers. No financing, no transaction.
• 8-K filed 2026-07-07 (event 2026-07-02): management changes only. Chris Sambar appointed Chief Enterprise Officer effective no later than 2026-10-14, reporting to the chief executive; André Almeida moves from Chief Broadband, Enterprise and Emerging Business Officer to Chief Marketing, Brand, and Broadband Officer; Dr John Saw's Chief Technology Officer remit expands to product engineering and cybersecurity; and Michael J. Katz, Chief Business and Product Officer, after more than two decades with the company, stepped away effective 2026-07-08, remaining a strategic adviser through December 2026. The press release is Exhibit 99.1 and is not present in the extraction.
• 8-K filed 2026-07-23 (event 2026-07-23): the second-quarter earnings release, furnished under Item 2.02. The primary document is a cover page. Exhibits 99.1 (press release) and 99.2 (Investor Factbook) are NOT in the archive, and the file contains no financial figures whatsoever. This is exactly the extraction limit the data contract warns about, and we record it as a finding rather than estimating around it: the company's own second-quarter release, including any 2026 guidance, is not available to this dive.

What the 8-K cover pages DO give us, and it is useful: the list of securities registered under Section 12(b) enumerates twelve separate exchange-listed senior note series — 3.550% due 2029; 3.700%, 3.150% and 3.200% due 2032; 3.625% due 2035; 3.850% due 2036; 3.500% due 2037; 3.900% due 2038; 3.800% due 2045; 6.250% due 2069; and two 5.500% series due March 2070 and June 2070. A 2070 maturity on an exchange-listed note is a fact about this capital structure that no vendor field carries.

Ownership and control

As of 2026-02-06, Deutsche Telekom AG held, directly or indirectly, approximately 52.8% of the outstanding T-Mobile common stock, and — through the Proxy, Lock-Up and ROFR Agreement with SoftBank dated 2020-04-01 — had voting control over approximately 56.9%. SoftBank ceased to be a related person on 2025-08-06. The 10-K's risk factors state plainly that "DT controls a majority of the voting power of our common stock and the T-Mobile trademarks we utilize in our business and may have interests that differ from the interests of our other stockholders."

This is a controlled company. Minority shareholders do not decide anything, and the T-Mobile brand itself is licensed from the controlling shareholder. Two shareholder derivative actions naming DT are outstanding — Dinkevich v. Deutsche Telekom AG (2021, over the Sprint merger repricing) and Palkon v. Deutsche Telekom AG (2025, over the 2022 Stock Repurchase Program and the 2023-2024 Stockholder Return Program).

2. The single most important thing to understand about these numbers: the metric that disappeared

Buried in the Performance Measures section of the June 10-Q, in one sentence with no emphasis, is the most consequential disclosure in this file:

> "Beginning with the three months ended March 31, 2026, we shifted away from reporting customer performance measures to better align with the Company's long-held priority on growing high-value accounts, which management believes is the best reflection of value creation versus customers."

T-Mobile has stopped reporting postpaid phone net customer additions.

That number is present throughout the FY2025 10-K. It is discussed by name — "Higher postpaid phone net customer additions, primarily from higher gross additions, partially offset by higher churn and increased deactivations from a growing customer base" — alongside postpaid phone churn ("increased 7 basis points, primarily due to higher industry switching") and postpaid phone ARPU ("increased $1.02, or 2%"). In the two 10-Qs filed since, it is gone. A grep of the June 10-Q for "postpaid phone" returns nothing.

What replaced it, and what the replacement says. The company now reports postpaid accounts, postpaid net account additions, postpaid account churn and postpaid ARPA. Here is what those show for the June quarter, verbatim:

> "Postpaid net account additions decreased 41,000, or 13%, for the three months ended June 30, 2026, primarily from: Higher account deactivations driven by the impact of a growing account base, including following the UScellular Acquisition, and higher average broadband-only accounts."

> "Postpaid account churn increased 7 basis points for the three months ended June 30, 2026, primarily from higher average broadband-only accounts, including following the acquisition of Metronet."

> "Postpaid account churn increased 9 basis points for the six months ended June 30, 2026, primarily from: Higher average broadband-only accounts... and Higher industry switching."

Three things follow and they are the whole bear case.

First, an account-level metric is structurally flattering and the company chose it. An account can contain many customers; ARPA rises when customers per account rises even if the company adds no customers at all. The 10-Q says exactly that: postpaid ARPA rose because of "An increase in customers per account, including from the continued adoption of 5G broadband and continued growth of T-Mobile for Business accounts." Moving from a customer metric to an account metric at the precise moment the customer metric turns is a choice, and we treat it as information about the customer metric.

Second, even the flattering metric is deteriorating. Net account additions down 13% in the quarter and 6% in the half; churn up 7 basis points in the quarter and 9 in the half; and one of the two named causes of the half-year churn increase is "Higher industry switching" — that is competitive, not mix.

Third, the fibre and broadband push is itself a churn source. The filing names "higher average broadband-only accounts" as the primary driver of rising churn in both periods. The growth initiative and the retention problem are the same accounts.

We are not saying the company is hiding a collapse. Postpaid revenue still grew 13% in the quarter, ARPA still rose $3.04, and Core Adjusted EBITDA still rose 12%. We are saying that the premium multiple T-Mobile carried was underwritten by a specific published number, that number is no longer published, and a 31.6% de-rating is a proportionate market response to that. Section 6 asks whether it is an over-response.

3. Growth and margin trajectory — and why the reported earnings understate the business

Annual

Read that table honestly and two facts stand out. The margin transformation happened in FY2023 — Sprint integration completing — and has now stalled: operating margin peaked at 22.1% in FY2024 and slipped to 21.2% in FY2025. And FY2025 net income FELL, from $11.339B to $10.992B, while revenue rose 8.5%. The 10-K's own summary: "Net income... was $11.0 billion and $11.3 billion for the years ended December 31, 2025 and 2024", with FY2025 including $293M net of tax of Workforce Transformation severance and $208M net of tax of impairment on capitalised software for the billing system ($278M pre-tax).

Also note the share count: 1,255.4M diluted in FY2022 down to 1,131.1M in FY2025 — a 9.9% reduction in three years, and 1,082.4M in the June 2026 quarter, a further 4.3%. The buyback is doing real work.

Quarterly — where the current picture actually is

The vendor's operating income reconciles to the filing exactly. The 10-Q states operating income "increased $277 million, or 5%, for the three months ended and was relatively flat for the six months ended June 30, 2026." Vendor: $5.490B less $5.213B = +$277M, +5.3%. Six months: $9.987B against $10.013B, −$26M. Both confirmed.

Now the point the brief asked us to test. The EPS sequence 1.88 → 2.27 → 2.99 looks like a violent operational ramp. It is not. Compare year on year instead of sequentially:

• Q4 2025 EPS $1.88 against Q4 2024's $2.56 — down 26.6%, on revenue up 11.3%. Operating income fell 15.7%.
• Q1 2026 EPS $2.27 against Q1 2025's $2.58 — down 12.0%, on revenue up 10.6%. Operating income fell 6.3%.
• Q2 2026 EPS $2.99 against Q2 2025's $2.84 — up 5.3%, on revenue up 7.9%. Operating income up 5.3%. Net income was FLAT: $3.239B against $3.222B, +0.5%. The entire EPS gain is the share count falling from 1,134.8M to 1,082.4M.

So the "ramp" is fourth-quarter seasonality reversing, plus a buyback. Fourth quarter is device-heavy and promotion-heavy; it is always the weakest earnings quarter. There is no discrete gain in the sequence — we looked for one and did not find one. The 10-Q's own summary of the half is the honest read: "Net income... was $5.7 billion and $6.2 billion for the six months ended June 30, 2026 and 2025, respectively." Reported net income fell 7.0%.

The clean-earnings correction — and it runs the other way

The reported decline is almost entirely disclosed, itemised, after-tax integration cost. The 10-Q lists them:

> "UScellular merger-related costs, including accelerated depreciation, net of tax, of $146 million and $622 million for the three and six months ended June 30, 2026, respectively, compared to $25 million and $35 million... for the three and six months ended June 30, 2025"

> "A gain related to the completed sale of a portion of our 3.45 GHz spectrum licenses, net of tax, of $113 million, for the three and six months ended June 30, 2025"

> "Costs associated with our Network Restructuring Initiative, including accelerated depreciation, net of tax, of $46 million and $149 million"

> "Severance and related costs associated with the 2025-2026 Workforce Transformation, net of tax, of $105 million for the six months ended June 30, 2026"

Reported first-half earnings fell 7.0%. Clean first-half earnings rose 8.6%. That 15.6-point gap is the single largest analytical fact in this dive, and it points the opposite way to the share price.

Clean trailing EPS. Trailing-twelve-month reported net income is $10.560B (Q3 2025 through Q2 2026), or $9.55 diluted EPS. Adding the $876M of first-half 2026 items and the $293M of FY2025 Workforce Transformation severance (which the 10-K places in the fourth quarter of 2025) gives at least $11.729B, or $10.84 per share — a 16.35x trailing multiple. Add the $208M software impairment, if it fell in the second half of 2025, and it is $11.937B, $11.03, and 16.07x. We call trailing clean EPS "at least $10.84" and the honest trailing multiple "about 16.1-16.4x", against the GAAP 18.56x. It is a floor, not a point estimate, because the second-half-2025 UScellular merger-related costs sit in an Adjusted EBITDA reconciliation table that the extraction stripped.

The caveat we owe the reader, stated plainly. T-Mobile has been adjusting out "merger-related costs" continuously since the Sprint merger closed in April 2020 — Sprint through mid-2024, then Ka'ena, then UScellular, with UScellular costs running to the end of fiscal 2027 and "additional restructuring initiatives" still under evaluation. A company that has excluded integration costs for six consecutive years and expects to keep doing so for two more is arguably running integration as an ordinary operating expense. We present both the reported and the clean numbers and we do not pretend the clean one is free.

Gross margin — rejected outright

The vendor prints grossProfitMarginTTM of 54.5%, FY2025 gross margin of 47.6% and FY2024 of 63.6%. All three are artefacts and none is used in this dive. The reason is Section 8, finding 4: the vendor moves depreciation and amortisation between cost of revenue and operating expenses from period to period. Reconstructed on a consistent depreciation-excluded basis, FY2025 gross margin was 62.9% against FY2024's 63.6% — a 70 basis point decline, not a 1,600 basis point collapse. Even that reconstruction we present only as a check, not as a headline.

Costs, in the company's own words

For the June quarter: cost of services up $261M or 10% on wholesale network access costs paid to Metronet and Lumos plus UScellular; cost of equipment sales up $396M or 8% on "Higher average cost per device sold, primarily driven by an increase in the high-end phone mix"; selling, general and administrative up $437M or 8% on UScellular, "Higher bad debt expense" and retail-initiative contract terminations ($108M pre-tax charge in the quarter); depreciation and amortisation up $288M or 9%, and up $907M or 14% for the half, on UScellular assets, the 5G build and accelerated depreciation from cell-site decommissioning. Interest expense, net up $133M or 14% in the quarter. Effective tax rate 25.2% in the quarter against 24.7%, and 25.0% for the half against 23.9% — a rising tax rate is a headwind consensus has to absorb.

4. Balance sheet and the enterprise-value rebuild

This is the section where the vendor is most wrong, and by the largest absolute amount in this batch.

What the vendor says (2025-12-31 — the most recent balance sheet in the payload)

What the filings say

The 10-K, at 2025-12-31: "our total debt and financing lease liabilities were $88.6 billion, excluding our tower obligations, of which $81.1 billion was classified as long-term debt and $1.1 billion was classified as long-term financing lease liabilities."

The 10-Q, at 2026-06-30: "our total debt and financing lease liabilities were $86.9 billion, excluding our tower obligations, of which $78.5 billion was classified as long-term debt and $1.1 billion was classified as long-term financing lease liabilities." And: "As of June 30, 2026, our Cash and cash equivalents were $2.8 billion compared to $5.6 billion at December 31, 2025."

Correction 1 — the vendor's capitalLeaseObligations of $35.987B is NOT finance leases; it is operating leases, and it does not belong in total debt. The 10-K says financing lease liabilities were $1.1B long-term. The vendor's short-term plus long-term debt of $86.282B, plus roughly $2.3B of current financing lease liabilities, is the filing's $88.6B. The $35.987B block — $4.977B current, $31.010B non-current — is the operating lease liability, which the vendor has mislabelled and folded into debt. The vendor reports total debt of $122.269B; the 10-K shows $88.6B of debt and financing leases; we use the filing.

Correction 2 — the balance sheet is six months stale and the 10-Q supersedes it. Between 2025-12-31 and 2026-06-30 T-Mobile issued $6.4B of long-term debt net and redeemed or repaid $7.8B of aggregate principal, and cash fell from $5.6B to $2.8B. The vendor's most recent balance sheet is 2025-12-31; the filing gives 2026-06-30; we use 2026-06-30.

Correction 3 — shortTermInvestments: 0 is CORRECT here, and this is worth saying. The standard defect (net debt ignoring short-term investments) does not apply to T-Mobile. Its balance sheet carries no short-term investment line; longTermInvestments of $2.765B is the equity-method interest in the Lumos and Metronet joint ventures, which is not cash and is correctly excluded from net debt. We checked, and the vendor is right on this one field. The error is elsewhere.

The rebuild

> Market cap $190.088B

> + total debt and financing lease liabilities $86.9B (10-Q, 2026-06-30)

> − cash and equivalents $2.8B (10-Q, 2026-06-30)

> = corrected enterprise value $274.19B

The vendor overstates enterprise value by $29.07B, or 9.6%. That is the largest absolute EV error we have found in this batch, and it is what makes T-Mobile screen as a 10.7x-EBITDA levered telco when it is an 8.3x one.

We present both the operating-lease-excluded and operating-lease-included figures and we do not force a choice. Including capitalised operating leases in telecom enterprise value is a defensible convention; the ~$36B figure is a 2025-12-31 number and no June-2026 equivalent is available in this file. What is not defensible is calling them capital leases and burying them inside totalDebt without a label, which is what the vendor does.

Correction 4 — trailing EBITDA, rebuilt from the 10-Q

The vendor's ebitdaMarginTTM of 30.72% on trailing revenue of $92.189B implies EBITDA of $28.318B. Against trailing operating income of $18.713B, that implies depreciation and amortisation of only $9.605B — for a company whose FY2025 cash-flow statement reports $13.508B of D&A.

The 10-Q settles it arithmetically. It states D&A "increased $288 million, or 9%, for the three months ended and increased $907 million, or 14%, for the six months ended June 30, 2026." The vendor's own first-half-2025 D&A is $3.198B + $3.146B = $6.344B; +$907M gives first-half 2026 of $7.251B; less the vendor's June-quarter $3.434B (which matches the filing's +$288M exactly) leaves $3.817B for the March 2026 quarter. The vendor's depreciationAndAmortization field for that quarter reads $1.150B — and the vendor's own otherExpenses field for the same quarter reads $3.817B. The vendor contradicts itself; the filing gives $3.817B; we use $3.817B.

Corrected trailing D&A: $3.408B + $3.756B + $3.817B + $3.434B = $14.415B.

Corrected trailing EBITDA: operating income $18.713B + D&A $14.415B = $33.128B (35.9% of revenue).

The vendor understates trailing EBITDA by $4.81B, or 17.0%.

Debt structure, covenants and headroom — what the filings disclose

• $10.0B revolving credit facility under the Second Amended and Restated Credit Agreement dated 2026-01-05, maturing 2031-01-05, with a $1.5B letter-of-credit sub-facility and a $500M swingline. No outstanding balance at 2026-06-30.
• Financial maintenance covenant of 4.5x on T-Mobile USA, Inc.'s Leverage Ratio. The 10-Q states: "We were in compliance with all restrictive debt covenants as of June 30, 2026." Our corrected 2.54x (or 3.62x with operating leases) is inside it, but the covenant uses the Credit Agreement's own definitions, which differ from ours; we report both and do not force a reconciliation.
• $2.0B unsecured short-term commercial paper programme — no outstanding balance at 2026-06-30.
• Effective interest rate, excluding derivatives and capitalised interest: 4.3% in the June 2026 quarter against 4.1% a year earlier, on weighted-average debt outstanding of $85.1B against $83.5B. For the half: 4.2% against 4.1% on $85.9B against $81.7B. Cost of debt is rising and the balance is rising.
• Instrument variety, from the 10-Q's fair-value note: Senior Notes to third parties, spectrum-backed Senior Secured Notes, Senior Notes to affiliates (i.e. to Deutsche Telekom), EUR-denominated Senior Notes (€7.3B, fully swapped to USD via cross-currency swaps designated as fair-value hedges), asset-backed notes ($2.0B at 2026-06-30, secured by $2.6B of gross equipment-instalment-plan receivables), Export Credit Agency facilities, and the Master Receivables Financing Agreement entered 2026-02-05, under which $1.0B is borrowed against pledged service customer relationships.
• Guarantor structure changed in the quarter. On 2026-03-31 T-Mobile USA elected to release the guarantees of certain subsidiaries; the obligor group is now T-Mobile USA as issuer plus T-Mobile, Sprint LLC, Sprint Capital Corporation and Sprint Communications LLC as guarantors. A guarantee release narrows the credit support behind the notes and is disclosed as a fact, not a judgement.
• Tower obligations are EXCLUDED from both the $86.9B and the $88.6B figures, and the balance sits in a table the extraction stripped. The structure is a failed-sale leaseback: approximately 6,200 CCI Lease Sites conveyed in 2012 and approximately 6,400 Master Lease Sites assumed with Sprint, recorded as financing obligations with the tower assets still on the balance sheet. CCI holds fixed-price purchase options totalling approximately $2.0B, exercisable per-tranche between 2035-12-31 and 2049-12-31. The 10-K's maturity schedule tail is "$3.2 billion in total thereafter."

The largest asset, and where the vendor put it

Roughly $100.7B of the $219.237B balance sheet is spectrum licences — derived from the vendor's own intangiblesToTotalAssetsTTM of 53.92%, which implies $118.21B of goodwill and intangibles, less goodwill of $13.678B and other intangibles of $3.843B. The vendor's intangibleAssets field reports $3.843B for FY2025 and $103.070B for FY2024, having moved the spectrum block into otherNonCurrentAssets (which jumps from $4.252B to $107.705B) in FY2025 only. The vendor's own ratio field and its own balance-sheet field disagree by $100B; we use the ratio field's implication and state that the balance-sheet line-item series is not comparable year to year.

And spectrum is impairment-tested qualitatively. The 10-K: "For our assessment of Spectrum license impairment, we employed a qualitative approach. No events or change in circumstances have occurred that indicate the fair value of the Spectrum licenses may be below their carrying amount at December 31, 2025." A qualitative test on the single largest asset in the company means an impairment, if it ever comes, arrives as a step change. Goodwill was likewise tested qualitatively, with the Wireless reporting unit's fair value estimated "using a market approach, which is based on market capitalization" — and market capitalisation has fallen 31.6% from the 52-week high since that test.

Equity is $59.203B against $219.237B of assets — 27.0%, with $30.545B of treasury stock. debtToEquityRatioTTM of 2.06x and priceToBookRatioTTM of 3.41x are arithmetically correct and economically thin, because equity is being deliberately consumed by buybacks. The vendor's own composite rating scores T-Mobile 1 out of 5 on debt-to-equity, dragging the overall grade to B / 3. We reject the debt-to-equity sub-score as a lens and use net debt to rebuilt EBITDA instead.

5. Cash flow and capital returns — and a free-cash-flow series that is not comparable

Two things in that table are traps and we walk the reader through both.

Trap one: the capex definition changes between years. In FY2022 and FY2025 the vendor's capitalExpenditure equals investmentsInPropertyPlantAndEquipment. In FY2023 and FY2024 it does not — it is $1.010B and $3.471B higher respectively, evidently including spectrum and intangible purchases. On the vendor's mixed basis, FY2025 free cash flow grew 80.3%. On a consistent property-and-equipment basis it grew 33.8%. Neither is right.

Trap two — and this is the one that matters: the 10-K says Adjusted Free Cash Flow grew SIX percent.

> "Adjusted Free Cash Flow increased $963 million, or 6%, for the year ended December 31, 2025."

And the reason for the divergence is disclosed in the same section:

> "Certain cash proceeds associated with the sale of receivables, which were recognized within investing cash flows before November 1, 2024, are recognized as operating cash flows. This change had no net impact to Adjusted Free Cash Flow."

That is the whole explanation. Operating cash flow rose $5.7B or 25% in FY2025 partly because receivable-sale proceeds moved into it from investing. The vendor's investing series corroborates: otherInvestingActivities was +$4.994B in FY2022 and +$3.979B in FY2023, then +$141M in FY2024 and −$4.129B in FY2025. The vendor reports free cash flow growing 80.3% in FY2025; the 10-K reports Adjusted Free Cash Flow growing 6%; we use the filing, and we do not use the vendor's multi-year free-cash-flow series to draw any trend conclusion.

The June 10-Q continues the pattern: "Adjusted Free Cash Flow increased $201 million, or 4%, for the three months ended and increased $404 million, or 4%, for the six months ended June 30, 2026." Four percent. Our arithmetic from those stated changes and (rounded) percentages implies roughly $5.2B in the June quarter and $10.5B in the half; because the percentages are rounded to whole numbers, those levels are approximate and are labelled as such.

First-half 2026 cash flows, verbatim from the 10-Q: operating cash flow up $883M or 6%; $5.3B of property and equipment purchases plus $510M of spectrum and intangible purchases; $7.8B of long-term debt repaid against $6.4B issued net; $7.1B of buybacks, $2.2B of dividends, $664M of financing lease repayments, $185M of tax withholding on share-based awards. Cash fell from $5.6B to $2.8B.

Capital returns — the swing factor, verified line by line

The dividend, and the vendor labelling trap the brief flagged. The vendor reports lastDividend: 3.94 and dividendPerShareTTM: 3.94, yielding 2.22%. That $3.94 is the trailing-twelve-month total, not the last declared quarterly dividend. The 10-Q gives the actual declarations:

> "On December 4, 2025, our Board of Directors declared a cash dividend of $1.02 per share... paid on March 12, 2026"

> "On March 19, 2026... a cash dividend of $1.02 per share... paid on June 11, 2026"

> "On June 15, 2026... a cash dividend of $1.02 per share, which will be paid on September 10, 2026"

Three quarters at $1.02 is $3.06; the trailing $3.94 implies the fourth quarter in the window was $0.88. The forward run-rate is $1.02 per quarter, $4.08 annualised — a 2.30% forward yield, not 2.22%. The dividend was raised roughly 16% between those two quarters.

The buyback, verbatim:

> "On December 11, 2025, we announced that our Board of Directors authorized our 2026 Stockholder Return Program of up to $14.6 billion... On April 23, 2026, we announced that our Board of Directors increased the 2026 Stockholder Return Program authorization to up to $18.2 billion."

> "During the three months ended June 30, 2026, we repurchased 11,420,845 shares at an average price per share of $188.76 for a total purchase price of $2.2 billion, and during the six months ended June 30, 2026, we repurchased 34,750,770 shares at an average price per share of $203.07 for a total purchase price of $7.1 billion."

> "Subsequent to June 30, 2026, from July 1, 2026, through July 17, 2026, we repurchased 2,149,600 shares at an average price per share of $182.53 for a total purchase price of $392 million... As of July 17, 2026, we had up to $8.5 billion remaining... through December 31, 2026."

Every tranche was bought above today's $177.21 — $203.07 for the half, $188.76 in the June quarter, $182.53 in July. That is a factual observation, not a criticism; it does mean the buyback has so far bought a falling stock.

The multi-year frame, also verbatim: "From January 1, 2026, through the end of 2027, the Company expects its business plan to support: Up to approximately $30.0 billion for share repurchases and cash dividends... and Over $22.0 billion in a discretionary and flexible envelope for opportunistic deployment, which may include de-levering, investments in our core business, strategic investments, and/or additional capital returns." This is management's own forward statement and is half-weighted accordingly.

Total shareholder yield, and whether free cash flow covers it

A 9.6% total shareholder yield is genuinely large and it is the strongest single argument for owning this stock.

But the coverage test the brief asked for does not pass cleanly. FY2025 free cash flow on the consistent basis was $17.995B; the 10-K's Adjusted Free Cash Flow is approximately $17.0B (our arithmetic from "increased $963 million, or 6%", and approximate because the percentage is rounded). Against $18.2B of intended returns, that is less than 1.0x coverage before a single dollar of strategic spending. And the strategic spending is committed:

• approximately $700M for the i3 Broadband joint venture, second half of 2026
• approximately $2.0B for the GoNetspeed/Greenlight joint venture, first half of 2027
• approximately $500M further capital contribution to Lumos, 2027-2028 — explicitly excluded from the 10-K's reported purchase-obligation table
• $420M Ka'ena earnout, second half of 2026
• $222M of Auction 113 spectrum, paid 2026-07-24 (after the balance-sheet date)
• $1.2B to $3.4B remaining under the Comcast 600 MHz agreement, targeted at the first half of 2028

Against that, one large inflow is pending: the Grain 800 MHz sale, $2.9B of cash consideration plus Grain's 600 MHz licences, FCC-approved 2026-07-01 and targeted to close in the third quarter of 2026, with "an increase to our cash income tax liability of approximately $850 million".

The honest conclusion: the 2026 capital return is funded by free cash flow plus asset sales plus a $2.8B cash balance that has already halved in six months. It is not fraudulent and it is not obviously unsustainable — the revolver is $10.0B and undrawn — but it is not self-funding either, and a reader should not treat a 9.6% shareholder yield on this balance sheet as equivalent to a 9.6% yield on an unlevered one.

6. Valuation — priced in or room?

At $177.21 (market cap $190.088B, approximately 1,072.7M shares, corrected enterprise value $274.19B):

Estimate coverage is good through 2028 — 13 to 18 analysts on the lines we use. The FY2029 EPS row rests on 11 analysts and FY2030 on FIVE; neither is used for any conclusion, and both carry an internal inconsistency (their netIncomeAvg divided by their epsAvg implies a share count of 1,131M, above the 1,013M implied for 2028 and above the 1,082M actually outstanding — the buyback simply stops in the model). Per the data contract, est.ebitdaAvg and est.ebitAvg are not used; we tested them and they fail — see Section 8, finding 8.

Basis note, and it is unusually clean here. The earn_cal actuals for the last three quarters — $1.88, $2.27, $2.99 — match inc_q GAAP diluted EPS exactly, and the FY2025 consensus of $9.895 landed 1.8% above the actual GAAP diluted $9.72. The estimate series is struck on a GAAP or near-GAAP basis, which is unusual and removes the GAAP-versus-adjusted ambiguity that distorts most large-cap forward multiples. All forward multiples in this dive are therefore GAAP-comparable, and the trailing multiple is shown on both the GAAP ($9.55, 18.56x) and clean (≥$10.84, ≤16.35x) bases.

Peer context. The vendor peer set is AMX ($74.7B), CMCSA ($88.5B), DIS ($170.5B), T ($160.2B), TDS ($3.8B), VIV ($19.7B) and VZ ($195.8B). Disney is not a comparable and is discarded. The relevant frame is that T-Mobile at $190.1B now sits BELOW Verizon at $195.8B for the first time in this file's history, and above AT&T at $160.2B. No peer earnings, multiples or estimates are supplied in the file, so no peer-multiple comparison is drawn — and we will not manufacture one from memory.

6a. What today's price assumes (the inversion)

At $177.21 — 12.75x the FY2027 consensus and 10.63x FY2028 — the price embeds the following falsifiable claims, each with a number and a date:

• Revenue grows only 6.9% in 2026, 4.4% in 2027 and 4.0% in 2028, to $102.550B. (Consensus-derived; 19, 18 and 16 analysts.) The market has already accepted deceleration. The nearest test is the 2026-10-22 print: consensus revenue of $23.157B would be +5.5% year on year and −4.3% sequentially. Note that the consensus EPS for that quarter, $2.86, is BELOW the $2.99 just delivered — the street is already modelling a sequential decline, so a simple beat of $2.86 proves very little.
• Net margin expands from 11.45% trailing to 14.98% in 2027 and 16.47% in 2028 — roughly 400 to 500 basis points. (Derived by us from the consensus revenue and net-income rows, which cross-check against epsAvg × implied shares to within 1% for 2026-2028.) This is the most fragile assumption in the price. It requires UScellular integration costs to roll off substantially in full — they were $622M net of tax in the first half of 2026 against $35M a year earlier — while the effective tax rate is rising (25.0% for the half against 23.9%) and interest expense is rising (up 13% for the half on a 4.2% average rate against 4.1%). The mechanism is real and disclosed: the company expects substantially all UScellular restructuring and integration costs by the end of fiscal 2027. But it is being asked to deliver 400-500 basis points of net margin on 4% revenue growth, which leaves no volume cushion.
• The share count falls from roughly 1,112M in 2026 to 1,013M in 2028 — 8.8% in two years. (Consensus-derived from netIncomeAvg ÷ epsAvg.) This is the most credible leg, because it is funded, board-authorised ($18.2B for 2026, up to approximately $30.0B across 2026-2027) and already being executed at 34.75M shares in the first half. Roughly one-quarter of the 2026-to-2028 EPS growth is this, and it is the leg least likely to fail.
• The market keeps paying at least 12.75x the forward year at every date between now and then. (Our number.) At 11x FY2028E the stock is $183; at 14x it is $233. The multiple, not the earnings, is the wider source of variance — and unlike the KLAC setup, this multiple has already compressed. Twelve months ago the same shares changed hands near $259, which on today's FY2027 estimate is 18.6x.
• And, implicitly: that the subscriber deterioration stops getting worse. (Our assumption on direction; the levels are filing-confirmed.) Postpaid net account additions down 13%, account churn up 7 basis points, prepaid revenue down 6%, wholesale down 8%. Nothing in a 12.75x multiple requires reacceleration. It does require the bleeding to stop.

6b. The return bridge (why the multiple moves)

Expected return over the next twelve months decomposes as: EPS growth (+27.4%, from FY2026E $10.904 to FY2027E $13.896) + multiple drift + dividend yield (+2.30%).

We are going to be pedantic about the multiple, because the guardrail matters here more than anywhere else in this batch. Today the market pays 12.75x the FY2027 estimate. In twelve months' time the forward year is FY2028. If the market then pays the same 12.75x on FY2028E of $16.669, the price is $212.6 — up 19.9% with no re-rating whatsoever. That is a rolldown outcome, not an expansion outcome, and we are not going to dress it up as one.

Our base of $208 is therefore, deliberately, slightly BELOW a constant-forward-multiple roll-forward. Expressed two ways, both disclosed:

• $208 = 12.5x the FY2028 consensus mean of $16.669 — a marginal de-rating from today's 12.75x forward multiple.
• $208 = 15.0x the FY2027 consensus mean of $13.896 — which, compared with today's 12.75x on the same year, looks like a 17.6% re-rating.

Both statements are true and they describe different clocks. The first is the honest one for a twelve-month horizon, and it is why our base assumes NO multiple expansion. Essentially all of the expected return is earnings growth plus the dividend.

The bull case at $259 likewise requires no re-rating at all — it holds the same 12.5x and applies it to the FY2028 consensus high of $20.684. That is deliberately conservative construction: the bull case is an earnings outcome, not a valuation outcome. The bear case at $131 is the only one that needs a de-rating, and it needs two things at once: the FY2028 consensus low of $14.606 and a compression to 9.0x. Say that plainly: on this name the upside does not need the crowd to change its mind, and the downside does. That asymmetry is the single most attractive feature of the setup, and it is why a stock this broken is a Watch rather than an Avoid.

6c. Variant perception (where we differ, what would surprise)

• We differ from the street on the number, and we are the bearish side of that difference. Consensus target is $235.50 with 44 buys, 9 holds and 1 sell across 54 analysts. Our base is $208 — 11.7% below. When our arithmetic lands materially below the crowd's on a stock that has fallen 31.6% and is still falling, the usual resolution is that the crowd's number comes down, not that the price goes up. The street's low target of $169 is already below spot. We treat the gap between $208 and $235.50 as unfinished business on their side, and it is a direct input to the Watch.
• We differ from the market on how to read the reported earnings decline, and this is our genuine edge on this name. Reported first-half net income fell 7.0%. Clean first-half net income, adding back $876M net of tax of company-disclosed and itemised UScellular integration, network-restructuring and severance charges, rose 8.6%. Clean trailing EPS is at least $10.84 against a GAAP $9.55, so the honest trailing multiple is about 16.1-16.4x, not 18.56x. If the market is de-rating T-Mobile on a headline profit decline that is entirely integration cost with a disclosed end-date of fiscal 2027, it is mispricing the name. Watchable number: UScellular merger-related costs net of tax in the 2026-10-22 print. They ran $146M in the June quarter against $476M in March. A continued step-down validates the clean lens; a step-up says integration is running long.
• We differ from every screen on the balance sheet, and by $29.1B. Every stock screener shows T-Mobile with $116.671B of net debt and 4.00x net-debt-to-EBITDA. The 10-Q shows $86.9B of total debt and financing leases against $2.8B of cash — $84.1B net — and rebuilt EBITDA of $33.128B, so 2.54x. Even including the ~$36B of operating leases the vendor mislabels, it is 3.62x, inside the disclosed 4.5x covenant. That moves T-Mobile from "dangerously levered telco" to "normally levered telco", and it is why the risk score is a 5 driven by the subscriber trend rather than a 3 driven by the debt.
• We do NOT differ on the growth question, and we say so. The market believes the growth premium is over. So do we, and the disclosure change is why. A company that stops publishing its headline growth metric in the quarter that metric turns has told you something, and no amount of clean-earnings arithmetic offsets it. We are not making a contrarian growth argument. We are making a cheapness argument on a decelerating asset with a very large capital return.
• Positive surprise that would force a re-rate: the October print showing postpaid net account additions and account churn stabilising, with UScellular merger-related costs stepping down again and the 2026 capital-return programme reaffirmed at $18.2B; or the Grain 800 MHz sale closing at $2.9B in the third quarter, which funds a chunk of the buyback without further debt; or a resumption of postpaid phone customer disclosure, which would be an unambiguous signal that management thinks the number is worth showing again.
• Negative surprise that would break the thesis: postpaid account churn rising again with "higher industry switching" named for a third consecutive period; any impairment against the ~$100.7B of qualitatively-tested spectrum licences or the $13.678B of goodwill (whose test uses market capitalisation, which has fallen 31.6%); UScellular integration costs re-accelerating or the completion date slipping past fiscal 2027; a cut to the 2026 stockholder return programme below $18.2B; or Deutsche Telekom disclosing an intention to sell, which the 10-K names as a specific risk.

Synthos fair values

All three anchors are multiples of the FY2028 consensus EPS distribution (mean $16.669, low $14.606, high $20.684, 13 analysts), cross-checked against FY2027E of $13.896 (17 analysts).

• Bear ~$131 — 9.0x the FY2028 consensus LOW of $14.606. Cross-check: 11.1x the FY2027 low of $11.816; and at $131 the $4.08 dividend yields 3.11%, which is not a yield floor for a US telco, so this scenario has room beneath it. The scenario: postpaid account losses accelerate, prepaid and wholesale keep shrinking, UScellular integration runs past fiscal 2027, the 400-500 basis points of margin expansion does not arrive, the buyback is cut to protect the covenant, and the multiple converges on a no-growth levered telco. This is the only one of the three cases that requires a de-rating. −26.1%.
• Base ~$208 — 12.5x the FY2028 consensus MEAN of $16.669. Cross-check: 15.0x FY2027E; corrected enterprise value at $208 would be $307.2B on FY2028E revenue of $102.550B, or 3.00x sales, against 2.97x today. Sensitivity, stated openly: 11.5x gives $192 — an 8% gain; 13.5x gives $225 — essentially the street's median of $237.50 less a haircut. The whole answer lives in an 11.5x-to-13.5x band and we claim no more precision than that. The scenario: revenue does its 4%, integration costs roll off roughly on schedule, net margin gets most but not all of the way to 16.5%, the buyback delivers its 8.8% share reduction, and the market pays roughly what it pays today. +17.4%, plus 2.30% of dividend.
• Bull ~$259 — the same 12.5x, applied to the FY2028 consensus HIGH of $20.684. Cross-check: 16.0x the FY2027 high of $16.206; and $259 is 0.02% below the 52-week high of $259.01 and 0.4% below the street's high target of $260 — three independent constructions landing within half a percent of each other, which is a coincidence worth noting rather than a confirmation. This bull case requires no re-rating, only the high estimate: churn stabilises, fixed-wireless and fibre convert to margin, integration ends early, and the discretionary $22.0B envelope is deployed into buybacks. +46.2%.

Base is 17.4% above spot; asymmetry roughly 1.77:1 to the upside (26.1% down, 46.2% up). That is a better ratio than the KLAC setup in this batch and a better one than most names carry at these prices. What makes it a Watch rather than a Buy is that our base sits 11.7% BELOW the street, the knowledge base is empty, the chart is below both moving averages with RSI at 42 — weak but not oversold, so there is no exhaustion signal — and there is no company-specific event for 79 days. We want this exposure. We do not want it into a downtrend with consensus targets still 12% above our own number.

7. Knowledge base — three claims, none about the equity, and three collisions inside a word

Raw hits: 6. Entity matches: 3. Text-only matches: 3. Used as supporting colour: 3. Used as name-level conviction: 0. Discarded for collision: 3.

The primary search covered TMUS, T-Mobile and TMobile across all 51,928 distilled claims. Supplementary sweeps on Mike Sievert, Sievert, Deutsche Telekom, UScellular, Metro by T-Mobile, Gopalan, fixed wireless and wireless carrier returned ZERO hits each — eight separate sweeps, nothing. Broad sector sweeps returned 5G: 59 text-only, 0 entity; telecom: 1 entity + 45 text-only; broadband: 1 entity + 14 text-only; Verizon: 1 entity + 4 text-only; AT&T: 6 entity + 6 text-only.

The three text-only "matches" on the ticker are pure homograph collisions and the mechanism is worth stating because it will recur. The search string TMUS is contained inside the word LITMUS. All three hits are sentences containing the phrase "litmus test" — one about Brazilian pension reform in 2019, one about US construction data in 2024, and one about a technology merger as a test of the deal environment in 2025. None mentions T-Mobile, telecommunications or wireless in any form. All three are discarded in full.

The sector sweeps are also discarded, for the reason the brief anticipated. Of the 59 "5G" text-only hits, none names T-Mobile; the top hit is a China-technology claim. The single "telecom" entity match is a 2020 Canadian-banks-and-REITs basket. The single "broadband" entity match is about Liberty Broadband's stake in Charter. Generic 5G, wireless and telecom claims are sector colour and are not admitted as name-level conviction. Of the six AT&T entity matches, none names T-Mobile — a claim about a competitor is not a claim about this company, and they are excluded on that ground.

That leaves three genuine entity matches. Here they are, verbatim.

> 2025-09-10 · bullish · conviction 55 · horizon: thesis · entities: Starlink, Verizon, AT&T, T-Mobile · channel: all_in · NO NAMED SPEAKER

> "Starlink could become a comprehensive global carrier (home Wi-Fi + direct-to-cell), though existing carriers won't be put out of business."

> 2025-03-18 · bullish · conviction 78 · horizon: thesis · entities: NVIDIA, Cisco, T-Mobile · channel: jensen_huang · NO NAMED SPEAKER

> "Nvidia/Cisco/T-Mobile will build a US radio-network stack; AI-RAN will accelerate the ~$100B/yr telecom capex market."

> 2026-03-16 · bullish · conviction 65 · horizon: thesis · entities: Nokia, T-Mobile · channel: jensen_huang · NO NAMED SPEAKER

> "The ~$2T telecom base-station infrastructure will be completely reinvented as AI infrastructure running AI at the edge (Aerial/AI-RAN)."

What this lane is, stated without inflation.

All three are entity matches, so T-Mobile genuinely appears in each entity list. Not one is a claim about T-Mobile as a security.

• The 2025-09-10 claim is about Starlink, with T-Mobile appearing as one of three incumbents being disrupted. Its stance is coded bullish, but bullish on Starlink — for T-Mobile it is a threat claim with a qualifier ("existing carriers won't be put out of business"). Read as a T-Mobile claim it is mildly negative, and it is the only independent voice in the lane.
• The 2025-03-18 and 2026-03-16 claims are AI-RAN infrastructure claims sourced to a channel slugged for the chief executive of a network-equipment supplier. Both predict that the telecom capital-expenditure market accelerates. For a supplier, telecom capital expenditure is revenue. For T-Mobile, it is a cost. These are promoter-class claims with respect to exactly the spending they forecast, and the house rule on promoter-class voices applies with the frame disclosed: we weight them low and we note that a claim predicting carriers will spend $100B a year more is not a bullish claim about carrier margins.

All three carry speaker: null and are attributed only to a channel. Under the 4-lane attribution policy that is the weakest form of sourcing we accept — a channel-level attribution tells you where a claim was said, not who said it or what their record is. No claim in the lane carries a management speaker_role, so the half-weight management discount does not apply; but two of the three carry the supplier's-management problem, which is functionally the same discount applied to a different company's executive.

Two adjacent claims are worth a single sentence each as colour, clearly labelled as NOT about T-Mobile. A 2023-03-13 claim (channel real_vision, no named speaker, entities Reliance Industries, Starlink, Huawei, AT&T) argues that "Base-infrastructure layer (5G/6G, Starlink) is going exponential; indebted telcos can't build, so Starlink owns data from space outside sovereign control" — a bearish structural claim about levered carriers that describes T-Mobile's category without naming it. And a 2025-03-13 claim (channel raoul_pal, no named speaker, entities Straight Path, AT&T) argues that "Wireless spectrum has consistently appreciated; bought at auction and held, it has delivered double-digit annualized returns" — relevant to the ~$100.7B of spectrum on this balance sheet, and the only independent corroboration in the file for the qualitative impairment test. Neither names T-Mobile; neither is admitted to the conviction pool; both are reported because a reader deserves to know they exist.

Conclusion, stated as a finding rather than dressed up. The Synthos knowledge base has no view on T-Mobile US, Inc. It has one satellite-disruption claim in which T-Mobile is an object rather than a subject, and two supplier-sourced capital-expenditure claims. None is name-level conviction, none justifies a conviction premium or discount in the fair value, and the absence — across eight zero-hit targeted sweeps including the current and former chief executive, the controlling shareholder and the largest acquisition — is itself the finding. Conviction rating Low, breadth 3, net conviction none, and the emptiness of the lane is a direct input to the Watch verdict.

8. Data integrity — what we rejected from the vendor file and why

Fourteen findings. Listed rather than silently corrected. T-Mobile is, on balance, one of the WORST vendor files in this batch on the balance sheet and cash flow, and one of the cleanest on revenue disaggregation. Both facts are stated.

1. bal_a is SIX MONTHS STALE and the 10-Q supersedes it — the largest single correction in this dive. The most recent balance sheet in the payload is 2025-12-31. The 10-Q filed 2026-07-23 gives 2026-06-30, during which T-Mobile issued $6.4B of long-term debt net, repaid $7.8B of aggregate principal, and let cash fall from $5.6B to $2.8B. The vendor's balance sheet is dated 2025-12-31; the filing gives 2026-06-30; we use 2026-06-30 throughout. We also scanned all three 8-Ks specifically for a financing that post-dates every statement in the file, per the defect-class-1 protocol. There is none — the three 8-Ks are an annual meeting, a set of management changes and an earnings furnishing. That check was run and it came back clean; the staleness problem here is ordinary lag, not a missed deal.

2. capitalLeaseObligations of $35.987B is the OPERATING lease liability, mislabelled and folded into totalDebt — REJECTED. The 10-K states financing lease liabilities of $1.1B long-term at 2025-12-31 and total debt and financing leases of $88.6B; the 10-Q states $1.1B and $86.9B at 2026-06-30. The vendor's $35.987B block ($4.977B current, $31.010B non-current) cannot be finance leases. The vendor reports total debt of $122.269B and net debt of $116.671B; the 10-Q shows $86.9B of debt and financing leases against $2.8B of cash, or $84.1B net; we use $84.1B. Consequential rejections: enterpriseValueTTM ($303.263B → $274.19B), netDebtToEBITDATTM (4.00x → 2.54x), evToSalesTTM (3.29x → 2.97x), evToEBITDATTM/enterpriseValueMultipleTTM (10.71x → 8.28x), evToOperatingCashFlowTTM (10.52x), evToFreeCashFlowTTM (19.16x). We also present the operating-lease-INCLUSIVE figures (EV $310.2B, 9.36x, 3.62x) because that convention is defensible in telecom; what is not defensible is the label.

3. Trailing EBITDA is understated by 17.0% because ONE quarter's depreciation figure is wrong — REJECTED and rebuilt. ebitdaMarginTTM of 30.72% implies trailing EBITDA of $28.318B, which against trailing operating income of $18.713B implies D&A of $9.605B for a company reporting $13.508B in FY2025. The 10-Q's stated first-half D&A increase of "$907 million, or 14%" on the vendor's own first-half-2025 base of $6.344B implies first-half-2026 D&A of $7.251B; less the June quarter's $3.434B (which matches the filing's stated +$288M exactly) leaves $3.817B for the March 2026 quarter — against the vendor's depreciationAndAmortization field of $1.150B and the vendor's own otherExpenses field of $3.817B for the same quarter. The vendor contradicts itself; we use $3.817B. Rebuilt trailing EBITDA = operating income $18.713B + D&A $14.415B = $33.128B (35.9% margin), and every EV/EBITDA figure in this dive is computed on it.

4. Gross profit and gross margin are NOT comparable across periods — REJECTED ENTIRELY, in every period. The vendor moves depreciation and amortisation between costOfRevenue and operatingExpenses from quarter to quarter and from year to year. Annually: FY2025 costOfRevenue is $46.241B (D&A inside) against FY2024's $29.653B (D&A in otherExpenses), manufacturing a fall in gross margin from 63.6% to 47.6% that did not happen. Quarterly, the same field reads $7.400B, $7.376B, $11.134B, $13.996B, $8.827B, $8.030B across the last six quarters — a gross margin swinging between 42% and 65%. On a consistent depreciation-excluded basis FY2025 gross margin was 62.9% against FY2024's 63.6%. No gross-margin figure from this vendor file appears as a headline anywhere in this dive.

5. The free-cash-flow series is not comparable across years, and the filing contradicts it outright. capitalExpenditure equals investmentsInPropertyPlantAndEquipment in FY2022 and FY2025 but exceeds it by $1.010B in FY2023 and $3.471B in FY2024. On the vendor's mixed basis FY2025 free cash flow grew 80.3%; on a consistent basis 33.8%; and the 10-K says "Adjusted Free Cash Flow increased $963 million, or 6%." The 10-K explains the divergence itself: "Certain cash proceeds associated with the sale of receivables, which were recognized within investing cash flows before November 1, 2024, are recognized as operating cash flows." The vendor's own otherInvestingActivities corroborates — +$4.994B (FY2022), +$3.979B (FY2023), +$141M (FY2024), −$4.129B (FY2025). The vendor implies 80% free-cash-flow growth; the filing says 6%; we use the filing and draw no trend conclusion from the vendor series.

6. lastDividend: 3.94 and dividendPerShareTTM: 3.94 are the TTM total, not the last declared dividend — corrected from the filings. The 10-Q records quarterly cash dividends of $1.02 per share declared 2025-12-04, 2026-03-19 and 2026-06-15. Three at $1.02 plus one at $0.88 is $3.94 — the vendor's figure, correctly computed but mislabelled. The forward run-rate is $4.08, a 2.30% yield, not 2.22%. dividendPayoutRatioTTM of 41.1% is computed on the trailing figure and is likewise a lagging number.

7. seg_prod is the revenue-DISAGGREGATION table, not the segment table — and here the distinction is definitive, because there IS no multi-line segment table. The 10-K, Note 1: "We operate as a single operating segment." The 10-Q's Note 11 – Segment Reporting presents "our Wireless segment", singular. The five seg_prod lines are the Note 11 revenue disaggregation from Revenue from Contracts with Customers. We verified the sums: FY2025 $88,309M, FY2024 $81,400M, FY2023 $78,558M and FY2022 $79,571M — all four EXACTLY equal to reported revenue, to the dollar — and every line change matches the 10-K MD&A's stated deltas and percentages. This is a clean field, used, correctly labelled.

8. The FY2021 seg_prod row does not sum — DISCARDED. Five lines total $77.795B against reported FY2021 revenue of $80.118B, short by $2.323B (2.9%). The FY2020 and FY2019 rows contain a sixth line, "Roaming and Other Service Revenue" ($2.078B in FY2020), absent from FY2021. The FY2021 row is incomplete and no conclusion in this dive touches it.

9. est.ebitAvg and est.ebitdaAvg are unreliable — TESTED against actuals, then NOT USED. The FY2025 ebitAvg of $13.607B is 27.4% BELOW the actual operating income of $18.744B; the FY2025 ebitdaAvg of $28.604B is 11.3% below our computed FY2025 EBITDA of $32.252B (operating income $18.744B + D&A $13.508B). Neither trips the standard detection signatures — ebitdaAvg exceeds ebitAvg in every year and neither is negative — so this file would pass the automated screen while still being wrong by a quarter. All forward valuation in this dive runs on epsAvg and revenueAvg.

10. The FY2029 and FY2030 estimate rows are internally inconsistent — EXCLUDED from every conclusion. netIncomeAvg ÷ epsAvg implies a share count of 1,131M in both 2029 and 2030, against 1,013M implied for 2028, 1,063M for 2027 and an actual 1,082.4M diluted in the June 2026 quarter. The model has the buyback stopping and the share count rising 12% — which is the opposite of the $18.2B programme actually running. Coverage is also thin: 11 analysts on FY2029 EPS and FIVE on FY2030. Both rows are reported for completeness and used for nothing.

11. intangibleAssets collapses from $103.070B to $3.843B between FY2024 and FY2025 while otherNonCurrentAssets rises from $4.252B to $107.705B — the year-on-year asset-composition comparison is REJECTED. T-Mobile did not dispose of $99B of spectrum; the vendor reclassified it into "other". The vendor's own intangiblesToTotalAssetsTTM of 53.92% implies $118.21B of goodwill and intangibles, of which roughly $100.7B is spectrum licences — and that ratio field is right while the balance-sheet field is wrong. We use the ratio field's implication and state that the line-item series is not comparable.

12. earn_cal and inc_q disagree on the September 2025 quarter — flagged, and inc_q used. The earnings calendar reports epsActual of $2.59 for the 2025-10-23 report; inc_q reports diluted EPS of $2.41 for the quarter ended 2025-09-30 ($2.714B net income ÷ 1,126.6M diluted shares = $2.409, internally consistent). The other five entries agree exactly — 2025-07-23 $2.84, 2026-02-11 $1.88, 2026-04-28 $2.27, 2026-07-23 $2.99 — all matching inc_q GAAP diluted EPS to the cent. We use $2.41 for the trailing-twelve-month calculation, giving trailing diluted EPS of $9.55 and a P/E of 18.56x, which is what the vendor's own priceToEarningsDilutedRatioTTM of 18.556 independently confirms. Using $2.59 would give $9.73 and 18.21x; the difference is not material to any conclusion, and the discrepancy is reported rather than resolved.

13. quote.yearHigh/yearLow disagree with tech.hi52/lo52 — we use tech and say so. The quote block reports a 52-week high of $261.56 and low of $165.66; the computed technical block reports $259.01 and $167.73. The discrepancies are 1.0% and 1.2%. We use the tech figures throughout, because they are computed from the same six-year close series that produces the moving averages, drawdown and relative-return figures. Note also that max_dd_from_peak of −35.05% exceeds pct_from_hi of −31.58%, implying a six-year peak of approximately $272.83 above the 52-week high of $259.01. Per the data contract that is NOT a defect and it is not reported as one.

14. Null and zero fields scanned; three found and explained. researchAndDevelopmentExpenses reads $0 for FY2020-FY2024 and $278M for FY2025 — T-Mobile does not disclose research and development as an income-statement line, so the FY2025 figure is a mapping artefact and the series carries no information; it is not used. generalAndAdministrativeExpenses and sellingAndMarketingExpenses read $0 for FY2021-FY2023 with only the combined SG&A populated, then split from FY2024 — the expense-composition series is not comparable and is not used. shortTermInvestments reads $0 in all three years and that is CORRECT (see Section 4, correction 3). seg_geo is an empty array — no geographic disaggregation is supplied at all, which is stated as a gap rather than estimated around.

Not defects, correctly reported and independently confirmed:

• Operating income matches the filing exactly — the 10-Q's "increased $277 million, or 5%" reconciles to $5.490B against $5.213B, and the first half's "relatively flat" to $9.987B against $10.013B.
• seg_prod sums to reported revenue to the dollar in four consecutive years, with every line change matching the 10-K MD&A's stated deltas.
• interestCoverageRatioTTM of 4.807x reconciles precisely to trailing operating income of $18.713B over trailing interest expense of $3.893B (we compute 4.81x).
• effectiveTaxRateTTM of 23.6% is consistent with the 10-Q's disclosed 25.2% for the quarter and 25.0% for the half against FY2025's 23.0%.
• netProfitMarginTTM of 11.45% reconciles exactly to trailing net income of $10.560B on trailing revenue of $92.189B.
• Currency: inc_a[0].reportedCurrency is USD and profile.currency is USD. CLEAN — no currency mixing, tested and confirmed.
• fullTimeEmployees of 75,000 matches the 10-K's "approximately 75,000 full-time and part-time employees" at 2025-12-31.

Share-count cross-check — run because the price level and the buyback demanded it, and it PASSES. Market capitalisation of $190,088,136,540 ÷ $177.21 implies 1,072.7M shares. Weighted-average diluted shares in the June quarter were 1,082.4M, and the ratios_ttm per-share fields imply 1,081.8M. The 0.9% gap between the market-cap count and the June-quarter diluted average is exactly what a live buyback produces: the company repurchased 11.42M shares during the quarter and a further 2.15M between 1 and 17 July, so a point-in-time count below the quarter's weighted average is arithmetically necessary. There is no share-class problem — there is a single class of common stock, par $0.00001, with 100 million authorised preferred shares and NONE outstanding at 2026-06-30. The one thing we could not verify directly: the 10-Q cover page's share count sits in a table the extraction stripped, so the cover-page figure is unavailable and we say so rather than implying we read it.

Non-equity tripwire — checked, one warning sign triggered, resolved. Beta is 0.319 — only marginally above the sub-0.3 threshold the tripwire flags, and on a name with a 2.30% dividend that deserves an explicit answer. It is common equity: the price of $177.21 is not par-like ($25, $50 or $1,000); the dividend is variable and was RAISED from $0.88 to $1.02 per quarter inside the trailing twelve months, which a fixed-rate instrument cannot do; the 52-week band of $167.73 to $259.01 is a 54% range, the opposite of the narrow band a fixed-income-like instrument shows; the 12-month price return is −25.3%, far outside any bond-like envelope; volume was 5,218,755 shares (roughly $925M of turnover); and the security is NASDAQ-listed common stock, par value $0.00001 per share, per the 8-K cover pages. The 0.319 beta is a genuine low-beta defensive-telecom characteristic, not an instrument-type error. This is common equity.

9. Technicals

• Price $177.21. −31.6% from the 52-week high of $259.01; +5.7% above the 52-week low of $167.73. Position within the annual range: 10.4th percentile — the bottom decile.
• Moving averages, both broken: −3.2% below a 50-day average of $183.04 and −10.7% below a 200-day average of $198.56. There is no supportive average anywhere above or below the price. The 50-day sits below the 200-day.
• RSI 42.2 — weak, and critically, NOT oversold. There is no mechanical exhaustion signal here. MACD −2.17 — negative.
• Maximum drawdown from the six-year peak: −35.0%, implying a peak of approximately $272.83. The current price is at, or within a fraction of, the worst point of that drawdown.
• Relative performance, and this is the section the brief said must not be softened: 3-month −8.9% against SPY +7.6% and QQQ +7.7% (a 16.4-point loss); 6-month −9.1% against SPY +11.1% and QQQ +15.7% (20.2 points); 12-month −25.3% against SPY +24.3% and QQQ +30.8% — a 49.5-point loss to the S&P and a 56.1-point loss to the Nasdaq 100 in twelve months. In a large-cap defensive telecommunications company, during a rising market, that is an extraordinary result and it is not a data artefact. It is the market repricing a growth stock as a value stock.
• Sentiment is the outlier in the whole file: 44 buy, 9 hold, 1 sell across 54 analysts, consensus target $235.50 (+32.9%), median $237.50, high $260 (+46.7%), low $169 — 4.6% BELOW spot. A 44-to-1 buy-to-sell ratio on a stock 31.6% off its high, with a consensus target 33% above the price, is not a bullish signal. It is an unrevised one. The gap between the street's $235.50 and our $208 is the size of the revision we think is still owed.

Today's move and what it does to the entry

TMUS closed 2026-08-04 at $177.21, up $0.12 or 0.07% from a previous close of $177.09. It opened at $175.22, traded a $174.20 to $177.33 range, and closed within 0.07% of the day's high on 5,218,755 shares against a 5,475,454 average. This was, functionally, an unchanged day on below-average volume.

And the context matters more than the move. 2026-08-04 was a violent sector-wide semiconductor rally — MRVL +12.8%, INTC +10.9%, LRCX +7.9%, MU +7.6%, KLAC +7.0%, AMD +7.0%, AMAT +5.5%, ASML +4.2%. T-Mobile participated in none of it, which is exactly what a 0.319-beta defensive telecommunications stock does on a risk-on day, and is itself a small piece of evidence that the instrument is behaving as advertised. No company-specific news appears in this file for 2026-08-04; the last company event was the 2026-07-23 earnings release, twelve days earlier.

The honest read on the entry, and it inverts the usual problem in this batch. Everywhere else in these twelve names the question is whether the reader is being asked to pay up for a one-day gap. Here there is no gap to pay for. The entry price is not distorted by a single session; it is the settled level after a twelve-month, 31.6% de-rating.

So the question is not "am I paying up today?" It is "has the selling stopped?" And the chart says no.

• The price is below both moving averages, the 50-day is below the 200-day, and MACD is negative.
• RSI at 42.2 is weak without being oversold — there is no capitulation reading and therefore no mean-reversion trigger.
• The stock is at the 10th percentile of its annual range, 5.7% above the 52-week low, in a −35.0% drawdown that has not yet made a higher low in this file.
• The company itself has been buying at $203.07, $188.76 and $182.53 — three successive tranches, each above the last close — and the price has kept falling. A $13.8B buyback that cannot hold a $190B stock is information.

What we are therefore telling the reader, plainly: the valuation case is good enough to want this and the chart is bad enough to wait for it. The two setups that would change it are the falsifiers:

• A hold of the 52-week low at $167.73, or a reclaim of the 50-day average at $183.04. The first would be the first evidence of a floor; the second the first evidence of a trend change. At $167.73 our base of $208 is +24.0% and the $4.08 dividend yields 2.43%.
• Or the 2026-10-22 print showing UScellular merger-related costs stepping down again with postpaid account churn flat or better, which would validate the clean-earnings lens directly. At that point paying today's price is defensible, because the uncertainty being priced has been resolved.

Buying today means catching a name that has lost 49 points to the index in twelve months, with no company catalyst for 79 days, and with a consensus target 12% above our own fair value that has not yet been marked down. That is not a trade with an edge in it — but it is a name worth a standing bid below.

10. Insiders — eight transactions, seven of them the annual director grant, and zero conviction either way

The reading, stated carefully. Every one of the eight transactions is coded A-Award — an acquisition by grant, not a purchase. Seven are the standard annual director equity award of 1,384 shares each, all dated 2026-06-16, which is the date of the annual meeting recorded in the 8-K filed 2026-06-18. That is a scheduled, formulaic, board-compensation event and carries no information. The eighth, 97.52 shares to the Chief Legal Officer at $187.02 on 2026-06-11, is a fractional accrual consistent with a dividend-equivalent or deferral credit, not a discretionary act.

What the file contains is therefore: ZERO open-market purchases and ZERO open-market sales. That is a neutral signal and it is reported as neutral.

Two observations are worth making anyway.

First, and mildly negative: nobody stepped in. The stock fell 31.6% from its high over this window and not one officer or director made a discretionary open-market purchase in the eight most recent filings. Discretionary buying into a drawdown is the single highest-information insider signal available, and its absence in a drawdown this deep is a small mark against.

Second, the residual holdings are asymmetric. Director Raul Marcelo Claure holds 1,849,801 shares — $327.8M at the closing price, an order of magnitude above every other named insider and above most of the executive team. Chief Legal Officer Mark Nelson holds 65,967. Several directors hold between 2,000 and 8,000 shares — less than $1.5M each, and in two cases under $500,000. A board with that little economic exposure to a controlled company in which the controlling shareholder holds 56.9% of the voting power is a governance observation, and we make it without drawing a conclusion from it.

Note the file contains no transactions from the chief executive or chief financial officer at all in the eight most recent records.

11. Verdict, kill-criteria and flip conditions

Watch.

This is the worst chart in the batch and the most-corrected balance sheet, and those two facts pull in opposite directions.

What has genuinely deteriorated, and none of it is in dispute: T-Mobile stopped reporting postpaid phone net customer additions beginning with the March 2026 quarter, and that number is present throughout the FY2025 10-K; postpaid net account additions fell 41,000, or 13%, in the June quarter; postpaid account churn rose 7 basis points in the quarter and 9 in the half, with "Higher industry switching" named as a cause; prepaid revenue fell 6%; wholesale and other service revenue fell 8%; reported first-half net income fell 7.0%; and the revenue growth that remains was bought — UScellular contributed 3,287,000 postpaid phone customers on 2025-08-01, Metronet 755,000 fibre customers, Lumos 97,000. Consensus has revenue growth decelerating to 4.4% next year and 3.3% by 2030. The premium multiple this company carried was underwritten by a published growth number that is no longer published, and a 31.6% de-rating and a 49-point twelve-month loss to the S&P is a proportionate response to that.

What is materially better than any screen shows, and this is why the name is a Watch and not an Avoid:

• Corrected net debt is $84.1B, not $116.671B, and corrected enterprise value is $274.19B, not $303.26B — the vendor mislabels ~$36B of operating leases as debt and runs off a six-month-stale balance sheet. Net debt to rebuilt EBITDA is 2.54x (3.62x including operating leases) against a disclosed 4.5x covenant the company confirms it is inside.
• Rebuilt trailing EBITDA is $33.128B, not $28.318B — the vendor's own quarterly depreciation field contradicts its own otherExpenses field, and the 10-Q settles it. EV/EBITDA is 8.28x, not 10.71x.
• Clean first-half net income ROSE 8.6% while the reported number fell 7.0%, on $876M net of tax of itemised, disclosed, end-dated integration and restructuring charges. Clean trailing EPS is at least $10.84 against $9.55 GAAP; the honest trailing multiple is about 16.1-16.4x.
• The capital return is $18.2B authorised for 2026 — 9.57% of the market capitalisation — with the declared dividend at $1.02 a quarter ($4.08 forward, 2.30% yield), up roughly 16% inside the trailing year, and up to approximately $30.0B across 2026-2027 plus a stated "over $22.0 billion" discretionary envelope.
• The bull case at $259 requires NO multiple expansion at all, only the FY2028 consensus high; only the bear case requires a de-rating. Asymmetry is 1.77:1.

What we are declining to pay for today: a base fair value of $208 that sits 11.7% BELOW the street's $235.50, on a name where 44 of 54 analysts still carry a buy; a chart below both moving averages with MACD negative and RSI at 42 — weak but not oversold, so there is no floor signal; 79 days with no company-specific catalyst; a knowledge-base lane of three claims, none about the equity, two from a supplier's chief executive and three discarded as collisions inside the word "litmus"; zero discretionary insider buying into a one-third drawdown; and a capital-return programme that is not covered by free cash flow once $5.0B to $7.2B of committed strategic outflows are counted.

The distinction that matters, and it answers the question the brief posed directly. The growth story HAS ended — the disclosure change proves it and the account and churn data confirm it. That is our answer and we are not hedging it. But the price has already fallen 31.6%, the multiple is 12.75x forward against a 9.6% shareholder yield, and the reported earnings decline that is driving the de-rating is a disclosed, itemised, end-dated integration cost sitting on top of clean earnings that grew 8.6%. A serious opportunity and a dead growth story are not mutually exclusive; that combination is precisely what a value name is. In the Synthos frame, a name where our fair value sits 11.7% below consensus, the conviction lane is empty and the trend is still down is a Watch by construction — we want the exposure, at a level, after the selling stops.

Pre-registered KILL criteria — what would take this to Avoid:

• Postpaid account churn rising again in the 2026-10-22 print, with "higher industry switching" named for a third consecutive period. That converts mix into competition and the multiple has to keep going.
• UScellular merger-related costs failing to step down, or the fiscal-2027 completion date slipping. The entire clean-earnings argument rests on those costs being finite; $146M in the June quarter against $476M in March is the trend that has to continue.
• Any impairment against the roughly $100.7B of spectrum licences or the $13.678B of goodwill. Both are tested qualitatively, and the goodwill test uses market capitalisation — which has fallen 31.6% since the last test date.
• A reduction in the 2026 Stockholder Return Program below the authorised $18.2B, or any language walking back the approximately $30.0B two-year framework. The shareholder yield is a third of the thesis.
• Third-quarter revenue below roughly $22.5B, or a full-year 2026 EPS trajectory implying below roughly $10.30 against the $10.904 consensus.
• A decisive break below the 52-week low of $167.73 on company-specific news rather than market beta.

Pre-registered FLIP conditions — what would take this to Buy — Tactical:

• A hold of $167.73 with a reclaim of the 50-day average at $183.04. This is the single most likely path to a Buy and it requires only a 3.3% rally after a base is made. At $167.73 the base case is +24.0% and the forward dividend yields 2.43%.
• The 2026-10-22 print with UScellular merger-related costs below roughly $100M net of tax and postpaid account churn flat or improving. Both legs of the clean-earnings variant perception validated at once, at which point paying 12.75x forward is straightforwardly cheap.
• The Grain 800 MHz sale closing at $2.9B in the third quarter, which funds a material slice of the buyback without further debt and demonstrates that the spectrum portfolio is monetisable at carrying-value-plus.
• Street consensus targets converging DOWN toward our $208 while the price holds. The current 11.7% gap resolving from their end rather than ours removes the main reason we are not buying.
• A resumption of postpaid phone customer disclosure. Management publishing that number again would be the strongest possible signal that it is worth publishing, and it would be worth roughly one tier on its own.
• Genuine name-level knowledge-base conviction appearing — a claim about T-Mobile as a security rather than as an object in someone else's thesis. At present there is none.

Where TMUS fits in the Synthos Framework Portfolio. The defensive-connectivity / high-shareholder-yield sleeve, at 0% today with a 2-3% target on a confirmed base near $168-172 or a clean October print. On the batch-overlap question: T-Mobile has almost no correlation with the semiconductor complex that dominates this batch — beta 0.319, a completely inert session on a day the semis moved 5-13%, and a 12-month return 49 points behind the index. That is genuine diversification rather than diversification dressed as concentration, and it is the strongest structural argument for eventually owning this name. Logged as a tracked Synthos call (Watch) as of 2026-08-04 at $177.21.

Single biggest risk: that the deceleration is structural and consensus 2027 EPS of $13.896 is wrong. Revenue is growing 4%. The EPS line needs 400 to 500 basis points of net-margin expansion and an 8.8% share-count reduction to reach $16.669 by 2028. The buyback leg is funded and near-certain. The margin leg is not: it requires integration costs to end on schedule while the effective tax rate is rising (25.0% for the half against 23.9%) and interest expense is rising (up 13% for the half on a 4.2% average rate against 4.1%). If the margin leg delivers only half, FY2028 EPS is nearer $14.60 than $16.67, and at 12.5x that is $183 — 3% above today's price. The forward multiple looks cheap because the E in it is doing a great deal of work.

Most fragile assumption in the price: that the subscriber deterioration stops getting worse. Nothing in a 12.75x multiple requires reacceleration — which is why the stock is not expensive. But it does require the bleeding to stop, and the company has just removed the metric that would let a reader check. Postpaid net account additions fell 13%, account churn rose 7 basis points, and the filing names "higher industry switching" and "higher average broadband-only accounts" — a competitive cause and a strategic one, both of which point at the next quarter as well as this one. We can no longer see postpaid phone net additions. Until the account-level numbers stabilise, we are being asked to underwrite a trend we cannot fully observe, and that is the reason this is a Watch.

Provenance & disclosures

• Traceability: 3 tagged knowledge-base claims name T-Mobile (raw hits 6, entity matches 3, text-only matches 3, discarded for collision 3; breadth 3, net conviction none). Sources: all_in (1 claim, 2025-09-10, bullish, conviction 55, no named speaker — a Starlink-disruption claim in which T-Mobile is an incumbent object rather than a subject) and jensen_huang (2 claims: 2025-03-18, bullish, conviction 78, no named speaker, on a US radio-network stack and AI-RAN accelerating telecom capital expenditure; and 2026-03-16, bullish, conviction 65, no named speaker, on base-station infrastructure being reinvented as edge AI infrastructure). All three carry speaker: null and are attributed only to a channel — the weakest sourcing the 4-lane attribution policy admits. NONE is a claim about T-Mobile as a security, and two are sourced to a network-equipment supplier's channel, making them promoter-class with respect to exactly the capital expenditure they forecast; they are weighted low with the frame disclosed. No claim in the lane carries a company-management speaker_role, so the half-weight management discount does not apply. The 3 text-only matches are discarded in full as homograph collisions: the search string TMUS is contained inside the word LITMUS, and all three hits are "litmus test" sentences — Brazilian pension reform (2019), US construction data (2024) and a technology-merger dealmaking test (2025) — none of which mentions T-Mobile, telecommunications or wireless. Searches: the primary sweep on TMUS, T-Mobile and TMobile across 51,928 distilled claims (6 hits, 3 usable); eight supplementary sweeps returning ZERO hits each on Mike Sievert, Sievert, Deutsche Telekom, UScellular, Metro by T-Mobile, Gopalan, fixed wireless and wireless carrier; and broad sector sweeps on 5G (59 text-only, 0 entity, none naming T-Mobile), telecom (1 entity — a 2020 Canadian-banks basket — plus 45 text-only), broadband (1 entity — Liberty Broadband/Charter — plus 14 text-only), Verizon (1 entity, the same all_in claim) and AT&T (6 entity, none naming T-Mobile, all excluded on the ground that a claim about a competitor is not a claim about this company). All quotes are verbatim from the stored claim text.
• Data as-of: income statement, cash flow and key metrics for fiscal 2025 ended 2025-12-31 plus nine quarters through 2026-06-30, from the vendor payload; the vendor's most recent BALANCE SHEET is 2025-12-31 and is superseded throughout this dive by the 10-Q filed 2026-07-23 for the period ended 2026-06-30 · estimates 2026-08-04 · prices 2026-08-04, quote timestamp 1785873601 = 2026-08-04T20:00:01Z ($177.21, +0.07%; 50-DMA $183.04; 200-DMA $198.56; RSI 42.2; MACD −2.17; 52-week range $167.73–$259.01) · knowledge-base claims 2026-08-04. Fiscal/calendar note: T-Mobile's fiscal year is the calendar year. All figures come from the Synthos vendor data file for TMUS or from the SEC filings in the TMUS archive; no figure comes from memory, recall or external retrieval.
• Filing archive contents and its gaps: 10-K filed 2026-02-11 (fiscal 2025); 10-Q filed 2026-04-28 (March 2026 quarter); 10-Q filed 2026-07-23 (June 2026 quarter — the freshest hard data in this dive); 8-K filed 2026-06-18 (annual meeting results, Item 5.07); 8-K filed 2026-07-07 (management changes, Item 5.02 — Chris Sambar appointed Chief Enterprise Officer effective no later than 2026-10-14; Michael J. Katz departing 2026-07-08 after two decades); 8-K filed 2026-07-23 (second-quarter results, Item 2.02). Known extraction limits, stated rather than estimated around: numeric tables are stripped from all extracted filing text, so the debt-maturity table, the Adjusted EBITDA reconciliation, the purchase-commitment table, the tower-obligation balance, the segment table, the earnings-per-share computation and the 10-Q cover-page share count are all UNAVAILABLE and none is estimated. The 2026-07-23 8-K is present as the cover document ONLY — Exhibits 99.1 (press release) and 99.2 (Investor Factbook), which carry the second-quarter financials and any 2026 guidance, were not extracted and the file contains no numbers whatsoever. The 2026-06-18 8-K's vote tallies are likewise absent. The 2026-07-07 8-K's Exhibit 99.1 is absent. No company guidance of any kind is available to this dive, and no guidance figure appears in it.
• Where the filings contradicted or corrected the vendor (detailed in Section 8): net debt rebuilt from $116.671B to $84.1B (the vendor folds roughly $36B of OPERATING lease liabilities into totalDebt under a capitalLeaseObligations label, and works off a 2025-12-31 balance sheet; the 10-Q states total debt and financing lease liabilities of $86.9B excluding tower obligations against cash of $2.8B at 2026-06-30), with enterprise value rebuilt from $303.263B to $274.19B and every EV multiple recomputed; trailing EBITDA recomputed from the vendor's implied $28.318B to $33.128B after the vendor's March-2026 depreciation field of $1.150B was shown to contradict both the 10-Q's stated first-half increase of "$907 million, or 14%" and the vendor's own otherExpenses field of $3.817B for the same quarter; the vendor's implied 80.3% FY2025 free-cash-flow growth REJECTED in favour of the 10-K's "Adjusted Free Cash Flow increased $963 million, or 6%", with the 10-K itself explaining the divergence as a reclassification of receivable-sale proceeds from investing to operating cash flow effective 2024-11-01; gross margin rejected entirely in every period because the vendor moves depreciation between cost of revenue and operating expenses (FY2025 47.6% against FY2024 63.6% is an artefact; on a consistent basis 62.9% against 63.6%); lastDividend of $3.94 identified as the TTM total rather than the last declared dividend, with the 10-Q giving three consecutive quarterly declarations of $1.02 and therefore a $4.08 forward run-rate and a 2.30% yield; the FY2021 seg_prod row discarded as $2.323B short of reported revenue; est.ebitAvg shown to be 27.4% below actual FY2025 operating income and est.ebitdaAvg 11.3% below our computed FY2025 EBITDA, so neither is used; the FY2029 and FY2030 estimate rows excluded for implying a share count of 1,131M against 1,013M for 2028; intangibleAssets shown to have been reclassified by roughly $99B between FY2024 and FY2025; quote.yearHigh/yearLow ($261.56/$165.66) rejected in favour of tech.hi52/lo52 ($259.01/$167.73) for internal consistency; and the earn_cal September-2025 EPS of $2.59 flagged against inc_q's $2.41, with $2.41 used. Where vendor and filing AGREED — worth recording: operating income to the dollar (June quarter +$277M, +5%; first half "relatively flat"); the seg_prod revenue disaggregation summing EXACTLY to reported revenue in FY2022, FY2023, FY2024 and FY2025 with every line change matching the 10-K MD&A's stated deltas; interestCoverageRatioTTM of 4.807x; netProfitMarginTTM of 11.45%; effectiveTaxRateTTM of 23.6% against the filing's 23.0% for FY2025 and 25.0% for the 2026 half; 75,000 employees; USD reporting currency against a USD quote; and shortTermInvestments: 0, which is CORRECT because T-Mobile's balance sheet carries no such line.
• Basis note: analyst estimates in this file appear to be struck on a GAAP or near-GAAP basis. The earn_cal actuals for the last three reported quarters — $1.88, $2.27 and $2.99 — match inc_q GAAP diluted EPS exactly, and the FY2025 consensus of $9.895 landed 1.8% above the actual GAAP diluted $9.72. All forward multiples in this dive are therefore GAAP-comparable. The trailing multiple is shown on both bases: 18.56x on GAAP diluted EPS of $9.55, and approximately 16.1-16.4x on clean EPS of at least $10.84 after adding back the company's own itemised, net-of-tax UScellular merger-related costs, Network Restructuring costs and Workforce Transformation severance. The clean figure is stated as a FLOOR, not a point estimate, because the second-half-2025 UScellular merger-related costs sit in an Adjusted EBITDA reconciliation table the extraction stripped.
• Estimate coverage: 19 analysts on FY2026 revenue and 17 on FY2026 EPS; 18 and 17 on FY2027; 16 and 13 on FY2028. The FY2029 EPS row rests on 11 analysts and FY2030 on FIVE; both are excluded from every conclusion, and both additionally carry an internally inconsistent implied share count. est.ebitdaAvg and est.ebitAvg were tested against FY2025 actuals, found 11.3% and 27.4% wrong respectively, and are not used.
• Peer note: the vendor peer set contains four broadly relevant names — Verizon ($195.75B), AT&T ($160.21B), Comcast ($88.47B) and América Móvil ($74.67B) — alongside Telephone and Data Systems ($3.80B), Telefônica Brasil ($19.72B) and The Walt Disney Company ($170.49B), the last of which is not a comparable and is discarded. No peer earnings, estimates or multiples are supplied in the file, so NO peer-multiple comparison is drawn anywhere in this dive. The one observation the file supports is that T-Mobile's $190.09B market capitalisation now sits BELOW Verizon's $195.75B.
• Fair-value caveat: the $131 / $208 / $259 anchors are multiples of the FY2028 consensus EPS distribution — 9.0x the low of $14.606, 12.5x the mean of $16.669, and the same 12.5x applied to the high of $20.684. Stated arithmetic, not a discounted cash flow. The base is deliberately sensitivity-disclosed: 11.5x gives $192 and 13.5x gives $225. The whole answer lives in an 11.5x-to-13.5x band and no greater precision is claimed. The bear case is the only one of the three that assumes a de-rating; the bull case assumes no re-rating at all, only the high estimate. Cross-checks: $208 is also 15.0x FY2027E; $259 is also 16.0x the FY2027 high and sits within 0.02% of the 52-week high of $259.01 and 0.4% of the street's high target of $260; $131 implies a 3.11% dividend yield on the $4.08 forward rate, which is not a floor for a US telecommunications company. Note explicitly that holding today's 12.75x forward multiple constant and rolling the calendar forward twelve months to FY2028E gives $212.6, so the $208 base assumes a marginal de-rating rather than any expansion.
• Timing: second-quarter 2026 results were released 2026-07-23, twelve days before this dive, and the 10-Q was filed the same day, so the freshest hard data is fully reflected. The next print is 2026-10-22, 79 days away (consensus revenue $23.157B, EPS $2.86 — below the $2.99 just delivered). There is no near-term company-specific catalyst, which is an explicit input to the Watch verdict rather than an oversight. 2026-08-04 was a violent sector-wide semiconductor up-day (MRVL +12.8%, INTC +10.9%, LRCX +7.9%, MU +7.6%, KLAC +7.0%, AMD +7.0%, AMAT +5.5%, ASML +4.2%); TMUS moved +0.07% on below-average volume and participated in none of it, which is consistent with its 0.319 beta and is treated as evidence of the instrument behaving as a defensive rather than as a signal.
• Controlled-company disclosure: Deutsche Telekom AG held approximately 52.8% of the outstanding common stock and voting control over approximately 56.9% as of 2026-02-06. The 10-K states that DT "may have interests that differ from the interests of our other stockholders", that DT controls the T-Mobile trademarks the company uses, and that future sales of stock by DT are a risk factor. Two shareholder derivative actions naming DT are outstanding. A reader should understand that minority holders do not control corporate outcomes at this company.
• Accessibility note: no information in this dive is conveyed by colour. All emphasis is carried by bold text, table structure and explicit labelling.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.