At+T Inc T

Thesis

AT&T is the textbook example of a 'great asset, mediocre business.' The company owns nationwide spectrum, fiber, and wireless infrastructure that connect roughly 120 million postpaid mobility customers and a growing fiber base — a tangible, real-economy business with predictable subscription revenue. The Buffett question is not whether AT&T will exist in ten years; it is whether the capital deployed inside it earns excess returns. The scorecard answers crisply: 10-year average ROIC is 6.65%, fcf_conversion_5y is 0.0, owner_earnings_ttm sits at -$21.3B, and the reverse-DCF/IV engine prints iv_base = -$25.97 and iv_high = -$25.97. A negative intrinsic value means the deterministic model believes capex plus interest plus dividends consume more cash than the operating business throws off on a normalized basis. P/E TTM of 15.92 vs a 10-year average of 11.35 also indicates the market is paying ABOVE history for an asset whose underlying economics have not improved. Net debt/EBITDA of -0.13 looks deceptively benign because the metric is distorted by lease accounting; reported gross debt remains well above $130B. The composite score of 63 reflects exactly this tension — balance sheet (16) and capital allocation (16) are passable, but profitability (14) is the constraint. At $26.12, the stock trades far above any positive IV the model can compute, so margin of safety is absent. The math says: until AT&T either (a) shrinks the asset base to free cash, or (b) the IV model recomputes positive on lower capex and stable ARPU, this is income, not compounding. Price/IV ratio is undefined because IV is negative — that is not a bug, it is the answer.

Moat

AT&T's moat is best described as 'narrow and structural, not durable and economic.' I work through the five moat archetypes individually.

1. Pricing power. Wireless service ARPU has been roughly flat for a decade and competes head-to-head with Verizon and T-Mobile. The Big Three have rough pricing parity; promotional intensity (free phones, family plans, $25/line MVNO offers from cable resellers) caps real pricing power. Damodaran's framing applies cleanly: when 'the firm is granted these rights by another entity, say the government, that entity usually preserves the right to control the prices charged and margins earned through regulation' [1]. Spectrum is licensed by the FCC, and the long-distance/wireless segment has historically traded pricing freedom for lower regulatory scrutiny. Verdict: weak.

2. Switching costs. Number portability (mandated in 2003) deliberately neutralized the strongest historical telecom moat. Damodaran's Microsoft Office discussion is the contrast: Office created compound switching cost by file-format lock-in and inter-app integration [3]. AT&T has nothing equivalent. Bundles (wireless + fiber discount, AT&T phone financing) create modest friction, and family plans add household-level inertia, but porting a number takes minutes. T-Mobile's net-add wins for years prove the switching cost is low. Verdict: narrow at best.

3. Network effects. Telecoms have engineering scale, not network effects in the Metcalfe sense. A larger subscriber base does not make the service more valuable to existing users — calls and texts work across networks. There is no winner-take-most dynamic here. Verdict: none.

4. Intangibles (brand, license, IP). This is where AT&T's real moat lives. Spectrum licenses are a quasi-legal monopoly: only three nationwide carriers hold the low-band, mid-band, and mmWave portfolio needed to operate. New entry requires a $50B+ spectrum-and-tower CapEx wall plus 10 years. This is the 'patents, licenses and other legal protection' bucket Damodaran discusses [2]. The brand carries a century of recognition but, per Damodaran [2], 'managers... who take over a valuable brand name and then dissipate its value will reduce the values of the firm substantially' — and AT&T's serial value-destroying M&A (DirecTV, Time Warner, Cricket) fits that pattern. Verdict: real but narrowing.

5. Cost advantages. Damodaran lists scale economies, exclusive distribution, and lower-cost inputs as the cost-advantage levers [3]. AT&T has scale (roughly $30B+ annual capex spread over 120M+ wireless subs), but T-Mobile after the Sprint merger has comparable scale at lower legacy-cost structure, and cable MVNOs (Comcast Xfinity Mobile, Charter Spectrum Mobile) ride the network at zero infrastructure cost — a structural cost-advantage erosion. Verdict: scale parity, no cost advantage over the relevant peer set.

$10B / 5-year stress test. If a well-funded entrant (a hyperscaler or a private-equity/sovereign consortium) deployed $10B over five years, they could not replicate AT&T's spectrum or tower footprint — that is the binding constraint. But $10B could fund a credible MVNO + fixed-wireless + satellite-direct-to-cell offering that takes 2-4 ppt of share in lower-ARPU segments. Starlink direct-to-cell with T-Mobile, and Amazon's Project Kuiper, are exactly this trajectory. The moat holds against frontal attack but leaks at the edges.

Erosion risk. Disruptive technology comes in slowly and from below per Damodaran [6]: 'the disruptive technology initially is targeted at small and less profitable markets and thus not viewed as a threat by established companies... The disruptive technology improves over time until it matches or even beats the dominant technology.' Direct-to-cell satellite is the textbook fit.

The persistent ROIC of 6.65% over a decade is the empirical verdict: a wide moat would show up as a 15%+ ROIC. The numbers do not lie.

Moat verdict: NARROW.

Management & Capital Allocation

Capital allocation at AT&T is the central question, because the scorecard's profitability (14) and capital_alloc (16) sub-scores reveal exactly where the value has historically leaked. I'll grade the five capital-allocation choices.

1. Reinvestment in the business. Reinvestment ROIIC is not meaningful in this period (scorer note: 'Net capital return period; ROIIC not meaningful') because the company has been net-returning capital and shrinking the asset base post-WarnerMedia spin. That is actually the correct response to a 6.65% 10-year ROIC: do not reinvest in a business that does not earn its cost of capital. The capex envelope ($20-22B/year) is being held flat in nominal terms while fiber is the marginal incremental dollar — that is the right call. The 5G/fiber buildout has at least a defensible reinvestment story; mid-band spectrum (C-band) deployed since 2022 is plausibly accretive.

2. Acquisitions. This is the historical disaster. DirecTV ($67B, 2015) and Time Warner ($85B + assumed debt, 2018) destroyed an estimated $40-60B of shareholder value when both were spun out at fractions of cost. Damodaran's brand-management warning applies precisely [2]: management 'who take over a valuable brand name and then dissipate its value, will reduce the values of the firm substantially.' Stankey (current CEO, formerly Stephenson's lieutenant) was inside the room for both deals. The current regime has been disciplined since the WBD spin (2022) — no major M&A — which is the right rehabilitation step but does not reverse the prior destruction.

3. Debt. Gross debt peaked over $180B post-Time Warner. The current trajectory is steady deleveraging — net debt/EBITDA reported at -0.13 (lease-adjusted GAAP, not economically meaningful given the leverage stack). Reported long-term debt remains north of $130B, and at current rate environment, ~$8B/year flows to interest expense. Refinancing risk is the main exposure: AT&T issued substantial low-coupon debt 2019-2021 (some 30-year notes at sub-4%); rolls in 2027-2030 will reprice meaningfully higher. Management has communicated a leverage target of ~2.5x net debt/adj EBITDA — credible.

4. Buybacks (and avg P/IV). AT&T resumed modest buybacks in 2024 after a multi-year pause for deleveraging. Authorization is $10B over multiple years. The critical Buffett test — buying back stock above intrinsic value is value destruction — is hard to evaluate when the model's IV is negative. Charitably, repurchases at $17-22 (2023-2024 range) may have been near book value; at $26 today, repurchases are arguably above any reasonable IV. Share count is up 1.7% over 10 years (share_count_change_10y = 0.017), meaning net dilution after a decade — buybacks have offset, not created, per-share value.

5. Dividends. The dividend was cut from $2.08 to $1.11 annual in 2022 alongside the WBD spin — this was a forced reset, not a Munger-clean policy change. Current $1.11 dividend at $26.12 yields ~4.25%, payout ratio roughly 60% of adjusted FCF. The cut was painful for retirees but mathematically correct: prior payout was unsustainable. Management deserves credit for taking the medicine.

Communication quality. Stankey's earnings calls are blunt and operationally specific (subscriber net adds, fiber pass-bys, capex pacing). No buzzword-laden narrative inflation. This is a positive. The 10-K is dense but transparent. Communication grade is solid B+.

Synthesis. History is D-grade (DirecTV, Time Warner). Current regime is B-grade (disciplined capex, deleveraging, painful but correct dividend reset). Forward-looking incentive structure (TSR-linked LTI) is reasonable but standard. Average over 10 years gives an honest C.

Capital allocator: C.

Industry Structure

U.S. wireless telecom is a structurally mature, regulated oligopoly. Porter's Five Forces:

1. Threat of new entrants — LOW (currently) / MEDIUM (10-year). Spectrum licensing, $50B+ network capex, and 10-year buildout requirements create an enormous moat against full-stack entrants. However, MVNOs (cable companies riding the host network) and satellite-direct-to-cell (Starlink, Kuiper) are entering at the edge with capex orders of magnitude lower. Damodaran's disruption pattern fits [6]: enter low-end, improve, then attack core. Comcast and Charter combined now have ~14M wireless lines, mostly stolen from the Big Three, including AT&T.

2. Bargaining power of suppliers — MEDIUM. Three handset OEMs (Apple, Samsung, Google) dominate, with Apple holding extraordinary leverage — iPhone subsidies are the largest single subscriber-acquisition line item. Network equipment is consolidated to Ericsson, Nokia, and (politically constrained) Huawei. Tower lease costs are increasing as American Tower, Crown Castle, and SBA exercise oligopoly pricing on their side of the table. Suppliers extract a meaningful share of telecom value pool.

3. Bargaining power of buyers — MEDIUM-HIGH. Consumers face zero search costs and full number portability. Promotional intensity (BOGO phones, $25/line cable plans) is structural. Enterprise customers run multi-carrier RFPs. Buyer power is real but bounded by the difficulty of switching all family lines simultaneously.

4. Threat of substitutes — MEDIUM and rising. WhatsApp, iMessage, FaceTime, and Zoom have already cannibalized voice and SMS revenue (the once-dominant profit pools). Wi-Fi calling reduces dependence on cellular for in-home use. Satellite direct-to-cell and fixed-wireless access (T-Mobile Home Internet has 5M+ subs) are substituting for both cellular and wired broadband. The substitute set is widening.

5. Rivalry among existing competitors — HIGH. T-Mobile took share aggressively post-Sprint merger and has not eased up. Verizon competes on premium positioning. AT&T sits in the middle with a fiber-bundle wedge. Promotional cadence (free phone trade-ins, line-of-credit financing) is constant. Margin compression is real, especially at the low end where cable MVNOs price at 50% of legacy carrier rates.

Value pool location and trajectory. Industry-wide service revenue grows GDP-ish (1-3%); equipment revenue is volatile. The growing pools are fiber broadband (capex-heavy but with real ARPU growth) and enterprise/IoT/private 5G. The shrinking pools are legacy voice, business landlines, and DirecTV satellite. AT&T's value pool is shifting from declining-but-cash-rich (legacy wireline) to growing-but-capex-heavy (fiber) — the timing of that crossover is everything.

Industry economics. ROIC industry-wide hovers at 6-9% (Verizon, AT&T) with T-Mobile somewhat higher post-merger synergies. This is below US large-cap median of ~14%. The capital intensity (capex/sales ~17%) is structurally high. There is no scenario where wireless telecom returns to the 15%+ ROIC business it was in the 1990s monopoly era.

Verdict. A profitable industry, but only barely above cost of capital, with rising substitute threats and persistent rivalry. Not a value-destroying industry, not a value-compounding one either.

Industry Verdict: Average.

Inversion (Bear Case)

I am now the short-seller. I am not hedging.

1. The single event that kills this. A regulatory + competitive double-shock. The FCC opens additional shared spectrum (CBRS expansion or new 12 GHz allocations) to cable, satellite, and private networks at low cost. Simultaneously, Starlink direct-to-cell scales from supplemental coverage to a credible primary service for the lower 30% of the ARPU distribution (~30-40M lines industry-wide). AT&T loses 8-12% of postpaid subscribers over 36 months while facing a $20-30B refinancing wall at materially higher rates. The combination — top-line decline + interest expense up $2-3B/year — pushes adjusted FCF below the dividend. The dividend is cut a second time within five years. The equity gets repriced as a melting ice cube, not a yield instrument.

2. Why the moat is narrower than bulls think. Bulls cite spectrum and scale. But the persistent 6.65% 10-year ROIC is the actual moat width — measured economically. A wide moat earns 15%+ ROIC for a decade. AT&T's spectrum is a moat against frontal-assault entrants only. It is not a moat against (a) cable MVNOs reselling network capacity at 50% of retail price, (b) satellite-direct-to-cell, (c) fixed-wireless as a substitute for fiber broadband, (d) WhatsApp/iMessage already having killed voice and SMS. The moat is a series of failing dikes, not a wall. The brand is also weaker than bulls think: NPS scores have trailed T-Mobile for years; brand recognition is not brand preference.

3. Why management is worse than it appears. The current regime is rehabilitating a 30-year capital-allocation track record so bad it would discredit any other CEO. DirecTV ($67B), Time Warner ($85B+), Vrio (Latin America DTV), and the failed AppleTV-style Mexican wireless venture were all approved by the same executive bench Stankey came up through. The 'discipline' bulls cite (no big M&A since 2022) is barely four years long against a multi-decade pattern. Compensation is still measured in adjusted EBITDA and adjusted FCF — both metrics that exclude the very cash items (capex, interest, restructuring) that distinguish accounting earnings from owner earnings. Owner_earnings_ttm is -$21.3B in the scorecard; if management were paid on owner earnings, the LTI plan would have paid zero. They are not. Stankey collected ~$24M in 2024 while owner earnings were negative.

4. What bulls are extrapolating that won't hold. Bulls extrapolate (a) fiber buildout reaching 30M+ pass-bys with 40% take-rate at $80 ARPU, (b) wireless ARPU growing 1-2%/year with stable churn, (c) tax tailwind on 100% bonus depreciation, (d) FedNow/IoT/private 5G adding $5B+ revenue by 2030. Each is fragile: cable competes for fiber take-rate at 60% of the price; T-Mobile fixed-wireless eats the rural and small-town customer segment fiber needs to make math work; bonus depreciation phases down statutorily; IoT and private 5G are tiny absolute revenue contributors with crowded competition (Microsoft, Amazon, Cisco, Ericsson). A more honest base case is flat-to-down service revenue 2026-2029, capex-light fiber that hits saturation by 2028, and continued share donation to T-Mobile and cable MVNOs.

5. Valuation trap (multiple compression / regime change). This is the cleanest part of the bear case. P/E TTM is 15.92 vs 10-year average of 11.35 — the stock is paying a premium multiple for a business with worse forward growth and worse balance-sheet flexibility than its 10-year average self. A reversion to the 10y avg multiple alone implies $26.12 × (11.35 / 15.92) = $18.62, a 29% drawdown with no operational deterioration. Layer in actual deterioration (5-10% revenue decline + interest expense up + dividend cut), and a 2x adjusted FCF reset to $13-15 is plausible. The IV model already prints negative ($-25.97 base, $-25.97 high, $-36.16 low); the math is telling you the reasonable range of intrinsic value is below zero on current capex-and-interest assumptions.

If I am right, the stock could be worth $13-15 within 3 years.

Lollapalooza Bias Check

Active biases in me, the analyst, right now:

1. Anchoring (active). I am anchored to AT&T's history as a widows-and-orphans dividend stock. The instinctive reaction is 'it's safe, it's a utility, it pays.' This anchoring biases me toward seeing the dividend as a floor rather than as an output of a capital-allocation calculation that was forcibly cut three years ago and could be cut again. The scorecard's negative IV explicitly contradicts the 'safe' framing.

2. Authority bias (active). Telecom is a 'serious' large-cap sector covered by every sell-side desk with consensus 'Hold' ratings and detailed multi-page DCFs. The presence of authoritative coverage creates a false sense that the price is 'about right.' But sell-side DCFs typically don't run a Buffett-style owner-earnings reverse-DCF — they use adjusted EBITDA multiples that the company itself constructs. Authority bias makes me weight institutional consensus more than the deterministic IV model in front of me; I should weight the model more.

3. Recency bias (active, in both directions). Recent positive: Stankey's disciplined post-WBD execution biases me toward grading current management higher than the 30-year track record warrants. Recent negative: T-Mobile's market-share gains bias me toward extrapolating share loss linearly. The honest answer is the trajectory is uncertain.

4. Commitment / consistency (mildly active). I have already drafted a fairly negative thesis above. There is a temptation to make the inversion section consistent with the bull case (a 'Hold') rather than the actual most-credible bear case (a 'Trim/Avoid'). I should resist softening.

5. Deprival super-reaction (mildly active). A 4.25% dividend yield is psychologically attractive in a 4% Treasury environment. The deprival reaction — fear of missing yield — biases me toward 'Hold' over 'Trim.' But the right comparison is yield vs total return; if the stock loses 20% of its principal to multiple compression, the dividend is irrelevant.

Inactive biases I should note. Social proof is mostly inactive (this is not a meme stock). Confirmation bias is partially active but checked by writing the inversion section. Incentive bias is low — I have no skin in the game.

Net effect. The biases are pulling me toward a softer call than the deterministic math justifies. The composite is 63 of 100. The IV is negative. The right call is the harder one.

10-Year Outlook

Same fundamental business model in 10 years? Yes — voice, data, video over wireless and fiber. Customer base larger? Roughly flat to mildly larger; US population growth is ~0.4%/year and wireless penetration is saturated, so any growth is share-driven. Profit per customer higher? Unlikely. ARPU has been flat for a decade across the industry; substitute services compress voice and SMS revenue; promotional intensity caps net pricing. The 'profit per customer' line trends sideways at best.

Moat wider in 10 years? No, narrower. Spectrum remains a barrier, but every adjacent technology — satellite direct-to-cell, cable MVNO, fixed-wireless access, Wi-Fi 7+, low-power IoT networks — chips at the edge. Disruptive technology improves slowly and from below [6]; the 10-year cumulative impact is large even if any single year's impact is small.

Single biggest threat? Satellite-direct-to-cell at consumer scale. Starlink + T-Mobile is already operational; Kuiper is 2-3 years out. If satellite achieves $20-30/month for a basic data-and-text plan covering 90%+ of the US population without towers, AT&T's lower-ARPU 30M+ subscriber tier becomes structurally vulnerable. This is the 'disruptive technology initially targeted at small and less profitable markets' [6] textbook playbook.

A Buffett-grade compounder requires a HIGH-confidence yes to 'will the moat be wider in 10 years?' AT&T cannot earn that yes. ROIC has been 6.65% over the last 10 years and there is no identifiable mechanism that lifts it above 10% over the next 10. The business will exist and will pay a dividend; it will not compound at attractive rates.

CONFIDENCE: low