AT&T Inc. (T)
T · New York Stock Exchange · Telecommunication services · Reporting currency USD · As of 7 August 2026
A cheap cash machine that must keep buying the spectrum its own free cash flow measure pretends is free.
Trigger to BUY: USD 20.50 or below, being the fair value midpoint of USD 27.37 less a 25% margin of safety. Six of eight quality tests pass. The two that fail are earnings quality and price, and only one of those can be fixed by waiting. Objective: income and modest total return. Holding period three to five years. Allocation band this instrument's characteristics support: 2% to 4% of a diversified equity portfolio.
| Metric | Value | Metric | Value |
|---|---|---|---|
| Market capitalisation | USD 162.5bn | Enterprise value | USD 310.7bn |
| Shares outstanding | 6.85bn | Free float | 99.9% |
| 52-week range | 19.89 – 29.79 | Trailing P/E | 7.8x |
| Forward P/E | 9.9x | EV / EBITDA | 6.8x |
| Dividend yield | 4.68% | Return on equity | 18.3% |
| Net debt, 30 Jun 2026 | USD 126.4bn | Beta, five year | 0.42 |
Source: AT&T second-quarter 2026 earnings release, 22 July 2026 (Tier 1); stockanalysis.com drawing on S&P Global Market Intelligence, close of 6 August 2026 (Tier 4). Enterprise value precedes the EchoStar spectrum payment that settled on 28 July 2026.
Why this matters now
- The USD 23bn EchoStar spectrum purchase closed on 28 July 2026, after the 30 June balance sheet date. Reported net debt of USD 126.4bn understates the position by about a fifth.
- On 4 August 2026 SpaceX said on its first earnings call that it will build a terrestrial network on 65 MHz of spectrum and take customers from carriers billing about USD 600bn a year. AT&T, Verizon and T-Mobile all fell.
- Second-quarter results on 22 July were strong enough to produce AT&T's largest weekly outperformance against the S&P 500 in twenty-five years. The stock is still down 14.6% over twelve months.
Three reasons to own it
- An 11% to 13% cash yield on equity. Management guides free cash flow of USD 18bn in 2026 rising to USD 21bn by 2028, against a USD 162.5bn equity value, before any re-rating.
- Convergence is working. 42.5% of advanced home internet households also buy AT&T wireless, and postpaid phone churn fell to 0.86% in the second quarter from 0.98% in the fourth quarter of 2025, despite price rises.
- The share count is finally falling. Down 2.15% over twelve months, with roughly USD 24bn of buybacks planned across 2026 to 2028 on a USD 162.5bn market value.
Three risks that matter
- A fourth national carrier. SpaceX bought the licence barrier rather than being blocked by it, which removes the ceiling that made US wireless a rational three-player market.
- Free cash flow excludes spectrum. USD 18bn guided for 2026 against USD 23bn paid for EchoStar licences in the same year, taking pro forma net debt to about USD 149bn and leverage to roughly 3.1x.
- Thin earnings quality. An 11.7% trailing effective tax rate against a 25% marginal rate, and a third of 2025 earnings per share from the DIRECTV disposal.
What the business actually is
AT&T sells connectivity on a monthly subscription. It says it serves more than 100 million US households and nearly 2.5 million businesses, and it is the third largest US wireless carrier behind T-Mobile and Verizon. Three products matter: a postpaid phone plan, a home internet connection delivered over fibre or fixed wireless, and business connectivity sold to enterprises and government. A fourth business, wireless in Mexico, is small and growing quickly.
Pricing is a rate card with periodic increases. Volume comes from net additions and from how many households take both the phone plan and the home connection. Handsets are sold alongside the service at close to cost, which inflates revenue without adding much profit: in the second quarter of 2026, service revenue of USD 25.9bn was 82% of the USD 31.5bn total and equipment sales were the other 18%.
Source: AT&T second-quarter 2026 earnings release and segment tables, 22 July 2026. Tier 1. Total operating revenues of USD 31,471m. Business connectivity combines business fibre and advanced connectivity of USD 1,946m with business transitional and other of USD 1,193m.
The 18% that is handsets is worth pausing on. Devices are resold at close to what they cost, so that line adds revenue and almost no profit. The other 82% is a bill that arrives every month, which is what makes this business defensive: households cancel a great many things before they cancel the phone.
Source: AT&T second-quarter 2026 earnings release, 22 July 2026. Tier 1. Segment EBITDA is operating income before depreciation and amortisation, as the company defines it.
Advanced Connectivity is 91% of revenue and 94% of segment EBITDA, and it is the only part growing. Legacy, the copper voice and data business, still produces USD 523m of quarterly EBITDA at a 32.0% margin, but revenue fell 25.9% year on year and management expects segment EBITDA to turn negative after 2027 until the direct costs of the copper network are eliminated. Part of current group EBITDA growth is arithmetic from shrinking a declining business, and the tail of that process is a cash cost rather than a benefit.
Inside Advanced Connectivity
The segment is not one business. In the second quarter of 2026 wireless service revenue of USD 17,413m grew 3.3%, advanced home internet of USD 2,926m grew 27.3% and business fibre of USD 1,946m grew 10.0%, while business transitional and other of USD 1,042m fell 16.6% and equipment of USD 5,137m was flat. Segment operating expenses fell 0.6% while revenue rose 4.1%, which is the whole reason segment operating income rose 20.3% on a single-digit revenue increase. That operating leverage is real, but it is the leverage of a business cutting cost faster than it grows, and there is a floor to that.
Earnings are moved by subscriber count, churn and the convergence rate. Postpaid phone average revenue per user was USD 56.57 in the fourth quarter of 2025 and fibre USD 72.87. In the second quarter of 2026 AT&T added 432,000 postpaid phone subscribers, plus 367,000 fibre and 279,000 fixed wireless connections. Customer concentration is nil, the natural advantage of a hundred million retail relationships. The real input dependency is spectrum, bought from the regulator at auction or from other licence holders, and that is the subject of the next section.
Competitive position and the shape of the moat
AT&T, Verizon and T-Mobile between them bill roughly USD 600bn a year, a figure given by SpaceX president Gwynne Shotwell on that company's first earnings call on 4 August 2026. Using a competitor's number rather than AT&T's own is deliberate, and the scale it describes is why the industry has been priced as a rational oligopoly since the Sprint merger. Share direction matters more than share level, and here AT&T reads well: more than 1.5 million postpaid phone subscribers added in each of the last five years, 432,000 in the second quarter of 2026, and the best year of consumer broadband growth in a decade during 2025.
| Company | Ticker | Mkt cap USDbn | P/E | Fwd P/E | EV/EBITDA | ROE | Debt/equity | 1-yr price |
|---|---|---|---|---|---|---|---|---|
| AT&T | T | 162.5 | 7.8x | 9.9x | 6.8x | 18.3% | 1.29 | −14.6% |
| Verizon | VZ | 195.2 | 12.2x | 9.3x | 8.1x | 15.8% | 1.84 | +10.1% |
| T-Mobile US | TMUS | 185.3 | 18.2x | 13.8x | 9.2x | 18.0% | 2.14 | −28.5% |
| Comcast | CMCSA | 89.3 | 4.6x | 6.8x | 4.8x | 20.9% | 1.07 | −26.9% |
| Charter | CHTR | 21.2 | 3.4x | 2.9x | 8.9x | 27.2% | 4.36 | −69.0% |
Source: stockanalysis.com drawing on S&P Global Market Intelligence. Tier 4. As-of dates differ: AT&T, Verizon and T-Mobile 6 to 7 August 2026, Charter 24 July 2026, Comcast 9 June 2026 for ratios, with market capitalisation at the close on 6 August 2026.
The peer set is a warning as much as a benchmark. Charter has lost 69% of its value in a year and Comcast 27%, both below five times earnings, because the market has decided a connectivity incumbent losing subscribers is worth a liquidation multiple rather than a utility multiple. T-Mobile, the share gainer, is down 28.5%. AT&T's 14.6% decline is the mildest in the group, which is the context for calling this stock cheap: it is cheap inside a sector the market has re-rated wholesale.
Barriers to entry, and the entrant who paid to cross them
The barriers are spectrum licences, sunk network capital and distribution. Morningstar records more than USD 60bn of licence spend by AT&T since 2020, and the company added USD 23bn in July 2026. Capital investment runs at USD 23bn to USD 24bn a year and fibre passes 38.6 million locations, targeted above 60 million by 2030. Replicating that is a decade of work.
EchoStar sold 65 MHz of spectrum to SpaceX for a combined USD 19.6bn and 50 MHz to AT&T for USD 23bn. SpaceX ended the second quarter of 2026 with USD 100bn of cash and 12 million Starlink subscribers, and on 4 August said it would attach small cellular base stations to Starlink dish mounts already on customers' buildings rather than build towers. A licence moat holds only while the licence is expensive relative to the entrant's balance sheet. For this entrant, it was not.
Pricing power, tested rather than asserted
The test is whether the company can raise prices through a cost cycle and keep the customer, and AT&T passes narrowly: it raised prices during 2026 and postpaid phone churn still improved to 0.86%. But wireless service revenue grew 3.3% while postpaid phone average revenue per user was broadly flat between the fourth quarters of 2024 and 2025, so most growth is coming from more customers rather than higher prices. Pricing power exists, it is modest, and it is the first thing a fourth national competitor would remove.
On regulation, the binding question is the copper wind-down. AT&T intends to stop serving most of its copper network by the end of 2029 and states in its own filings that California approval could push that date out. Delay is expensive, because the fixed cost of the estate is removed only once every customer in a geography has been migrated. On disruption, the honest answer is that satellite direct-to-device service was a partnership opportunity until this month and is now an announced competitive threat that nobody, including AT&T's management, can currently size.
The mechanism is a regulatory licence combined with sunk network capital and fibre passings, which together make replication slow and expensive. It is not wide, because the licence can be bought by anyone with a large enough balance sheet, and in the last twelve months somebody did exactly that.
Financial performance and the quality behind it
The series below starts in 2024 because the WarnerMedia separation in 2022, the DIRECTV disposal in the third quarter of 2025 and the reclassification of the acquired Lumen fibre network to discontinued operations in the first quarter of 2026 make earlier years incomparable on the current basis.
| FY24 | FY25 | FY26E | FY27E | FY28E | |
|---|---|---|---|---|---|
| Revenue | 122.3 | 125.6 | 128.5 | 131.1 | 134.0 |
| Adjusted EBITDA | 44.8 | 46.4 | 48.0 | 50.2 | 52.7 |
| Adjusted EBITDA margin | 36.6% | 36.9% | 37.4% | 38.3% | 39.3% |
| Adjusted EPS, USD | 1.95 | 2.12 | 2.30 | 2.56 | 2.86 |
| Capital investment | 22.1 | 22.0 | 23.5 | 23.5 | 23.5 |
| Free cash flow | 15.3 | 16.6 | 18.2 | 19.3 | 21.2 |
| Dividend per share, USD | 1.11 | 1.11 | 1.11 | 1.11 | 1.11 |
| Weighted diluted shares, bn | 7.18 | 7.20 | 6.95 | 6.68 | 6.43 |
| Net debt | 120.1 | 117.4 | 148.0 | 143.1 | 136.0 |
| Net debt / adjusted EBITDA | 2.68x | 2.53x | 3.08x | 2.85x | 2.58x |
History from AT&T's releases of 27 January 2025, 28 January 2026 and 22 July 2026. Tier 1. Weighted diluted shares are derived as net income attributable to common stock divided by reported diluted earnings per share. E denotes our estimates; every input is printed in the assumption block on the valuation section below.
Revenue 2021 to 2025 from stockanalysis.com drawing on S&P Global Market Intelligence (Tier 4), with 2025 cross-checked to the USD 125.6bn reported by the company. The 2022 fall is the WarnerMedia separation, not a trading decline. Forecast years are ours.
Quality of growth: how much of it was bought
Second-quarter revenue rose 2.3%, which the company attributes to fibre and wireless growth, to the Lumen mass markets fibre business acquired on 2 February 2026, and to favourable Mexican exchange rates. Management has quantified the acquired portion: Lumen retail fibre subscribers add roughly 100 basis points to Advanced Connectivity service revenue growth in 2026, so the 5.1% segment service revenue growth is nearer 4.1% organic. Still good, but reporting 5.1% without the split would flatter the business by a fifth of its growth rate.
Quality of earnings: four things the headline does not show
- A third of 2025 earnings was a disposal. Diluted earnings per share of USD 3.04 included USD 0.80 from the gain on the DIRECTV sale and USD 0.21 of DIRECTV equity income. Adjusted earnings per share for the same year was USD 2.12.
- 2024 carried a USD 4.4bn goodwill impairment. Reported operating income of USD 19.0bn against adjusted operating income of USD 24.2bn.
- The tax rate is temporarily tiny and it reverses. The trailing effective rate is 11.7% against a marginal rate the company itself puts at approximately 25%, with cash taxes guided at only USD 1.0bn to USD 1.5bn in 2026 under the One Big Beautiful Bill Act. That is a timing benefit worth roughly USD 3bn a year of free cash flow.
- A spectrum write-off in the quarter of a spectrum purchase. The second quarter of 2026 carried a USD 0.03 per share asset abandonment charge tied to a reprioritisation of the spectrum strategy, recognised in the same quarter the company was completing a USD 23bn spectrum acquisition.
None of this is improper and the disclosure is complete. All of it means the reported and adjusted figures have to be read together rather than separately.
Balance sheet, cash generation and what reaches the holder
The published balance sheet is dated 30 June 2026. The USD 23bn spectrum payment settled on 28 July. A reader using reported net debt of USD 126.4bn understates the position by about a fifth. Management's guidance already reflects the deals: leverage reaches roughly 3.2x after Lumen and EchoStar, falls to about 3.0x by the end of 2026, and returns to the 2.5x range within approximately three years of closing. Getting to 3.0x this year depends on selling a controlling interest in the acquired Lumen fibre subsidiary to an equity partner, which is an announced intention rather than a completed transaction.
Moody's affirmed AT&T at Baa2 on 27 August 2025 when the EchoStar deal was announced, expecting leverage back to 3.5x or better within two years of closing. The blended cost of that debt, derived as interest expense over average total debt, was 5.24% for 2025 and 5.43% on the second quarter of 2026 annualised. Interest cover is 3.6x. A USD 17.5bn delayed draw term facility closed in the fourth quarter of 2025 is how the spectrum was paid for.
| Measure | At USD 40bn trough adjusted EBITDA | Comment |
|---|---|---|
| Interest cover | 5.3x | USD 7.6bn of annual interest covered comfortably by EBITDA alone |
| Cover after capital investment | 2.2x | EBITDA less the USD 23.5bn capital programme, divided by interest |
| Net debt / adjusted EBITDA | 3.7x | From 3.1x pro forma. No covenant breach implied by disclosure |
| Implied free cash flow | USD 7.6bn | Against a dividend costing about USD 7.7bn. Exactly covered and nothing more |
Our calculation. The assumed trough is a 17% fall from our 2026 estimate of USD 48.0bn, a deeper decline than this business has produced in any recent cycle. Interest, capital investment and cash tax inputs are the company's own guided figures for 2026.
The company survives that scenario comfortably. What it cannot do at a trough is fund the dividend, the buyback and the next spectrum purchase at the same time, and that is the constraint a holder is actually taking.
Free cash flow, and what it leaves out
Free cash flow of USD 16.6bn in 2025 converted at 36% of adjusted EBITDA, after USD 20.8bn of capital expenditure and USD 1.2bn of vendor financing, with cash from operations of USD 40.3bn. Capital investment of USD 22.0bn was 17.5% of revenue, and the guided step up to USD 23bn to USD 24bn takes that to about 18%, a level the company says it can hold while free cash flow still rises to USD 21bn by 2028.
AT&T's free cash flow is cash from operations less capital expenditure and vendor financing. It does not include spectrum, which is an investing outflow. In 2026 the company will report roughly USD 18bn of free cash flow having spent USD 23bn on airwaves and USD 5.75bn on the Lumen fibre business. Net debt rises by around USD 30bn in a year described as generating USD 18bn of free cash. This is not a scandal and the disclosure is complete. It is simply that the headline number a great many holders rely on is not measuring what they think it measures.
2024 and 2025 are weighted average diluted shares derived from reported net income attributable to common stock and reported diluted earnings per share. 2026 to 2028 are our estimates, built from the roughly USD 24bn of buybacks management plans across the period.
The dividend has been held at USD 1.11 since the 2022 rebase, costs about USD 7.7bn a year, and is 37% of earnings and 42% of guided 2026 free cash flow, funded by operations rather than the balance sheet. Buybacks were USD 4.3bn in 2025 and are being accelerated to roughly USD 10bn in 2026. After years in which the count barely moved, it is falling at just over 2% a year, removing about 750 million shares between 2025 and 2028. That is the mechanism by which flat group earnings become growing per-share earnings, and the largest single contributor to the guided double-digit adjusted earnings per share compound rate.
Management, governance and capital allocation
The record against the targets management published
| Target, 2024 Analyst Day | Promised | Delivered or now guided | Verdict |
|---|---|---|---|
| Adjusted EPS 2025, ex-DIRECTV | USD 1.97 to 2.07 | USD 2.12 | Beaten |
| Free cash flow 2025, ex-DIRECTV | USD 16bn or better | USD 16.6bn | Met |
| Capital investment, 2025 to 2027 | About USD 22bn a year | USD 22.0bn in 2025 | Met |
| Free cash flow 2027 | USD 18bn or better | Now guided USD 19bn or better | Raised |
| Shareholder returns | USD 40bn, 2025 to 2027 | USD 45bn, 2026 to 2028 | Raised |
| Net debt / EBITDA through 2027 | Held near 2.5x | 3.2x after the two acquisitions | Abandoned |
| Total fibre locations | More than 50 million | 38.6m, targeting 60m by 2030 | On track |
Targets from AT&T's Analyst and Investor Day release of 3 December 2024, filed as a Form 8-K exhibit. Outcomes from the releases of 28 January 2026 and 22 July 2026. Tier 1 and Tier 2.
The operating record is good. Every consolidated financial target set in December 2024 for 2025 was met or beaten, and the outlook beyond has since been raised rather than trimmed. That is not the norm in this sector and it should be credited.
The commitment to hold net debt at around 2.5 times adjusted EBITDA through 2027 was dropped within fifteen months of being made, to fund USD 28.75bn of acquisitions in a single year: USD 5.75bn for Lumen's mass markets fibre business and USD 23bn for EchoStar spectrum. Both may prove good purchases. Neither was contemplated when the target was set. A management team that hits its operating numbers and moves its balance sheet targets to accommodate deals is telling shareholders which promise is load-bearing.
Capital allocation, where the cash went
In 2025 AT&T invested USD 22.0bn in the network, paid roughly USD 8bn of dividends, repurchased USD 4.3bn of stock and received about USD 5.4bn after tax from DIRECTV. In 2026 it has committed USD 28.75bn to two acquisitions and roughly USD 10bn to buybacks while the network programme steps up to USD 23bn to USD 24bn. The spectrum purchase is the item to judge management on in five years: USD 23bn buys 30 MHz of nationwide 3.45 GHz mid-band and 20 MHz of nationwide 600 MHz low-band, which is capacity rather than revenue, and its return shows up as churn that did not happen rather than as a line in the accounts.
Ownership and control
| Holder | Share |
|---|---|
| The Vanguard Group | 9.37% |
| BlackRock | 8.11% |
| State Street Global Advisors | 4.57% |
| Newport Trust | 2.59% |
| JPMorgan Chase | 2.56% |
| Geode Capital Management | 2.41% |
| Bank of America | 1.63% |
| Morgan Stanley | 1.55% |
| GQG Partners | 1.51% |
| Norges Bank | 1.24% |
| All institutions combined | 69.8% |
| Executives and directors combined | 0.11% |
Source: Form 13F filings compiled by Yahoo Finance, reported 30 September and 31 December 2025 (Tier 4). Institutional, insider and float percentages from stockanalysis.com drawing on S&P Global Market Intelligence, 6 August 2026 (Tier 4).
There is no controlling holder, no founding family, no government stake and no dual-class structure. Institutions hold 69.8% of the register and the ten largest holders about 35.5%, almost all index money that does not sell and rarely votes against a board. Executives and directors hold 0.11% between them, roughly one share in every nine hundred. That is normal for a company of this age and size, and it means alignment rests entirely on the incentive plan rather than on personal capital at risk. Free float is 99.9% of shares outstanding, so liquidity is excellent and there is no minority-treatment question of the kind that arises in family or state-controlled listings.
Valuation
Model selection. A free cash flow to the firm discounted cash flow, cross-checked on a mid-cycle EV to EBITDA multiple. We make one adjustment the company's own measure does not: spectrum is charged as a recurring capital cost rather than an occasional investing item, because a wireless carrier cannot stop buying capacity and remain one.
| Ratio | AT&T now | Own five-year | Peer median | Comms services |
|---|---|---|---|---|
| Trailing P/E | 7.8x | 9.0x | 8.4x | 20.5x |
| EV / EBITDA | 6.8x | not sourced | 8.5x | not sourced |
| Price / book | 1.5x | not sourced | 2.0x | not sourced |
| Free cash flow yield | 10.9% | not sourced | 11.0% | not sourced |
| Dividend yield | 4.68% | not sourced | 5.6% | not sourced |
Five-year median P/E of 9.0x from GuruFocus, 7 July 2026. Peer medians computed from the peer table in section 02, excluding AT&T. Sector column is the Communication Services weighted average P/E of 20.53x from stockanalysis.com. Tier 4. Where a comparable value could not be sourced on the same basis this run it is marked as such rather than estimated.
Assumption block - discount rate
- Risk-free rate 4.67%. US 10-year Treasury, 6 August 2026.
- Equity risk premium 5.5%. Our assumption, a long-run US premium.
- Beta 0.85 applied. The observed five-year beta is 0.42, which gives a 7.0% cost of equity, a 5.8% cost of capital and a fair value above USD 32. We reject it: it was measured while AT&T traded as a rate-sensitive income proxy and cannot capture 3.1 times leverage plus a national entrant announced this month.
- Cost of equity 9.35%. 4.67 + (0.85 × 5.5).
- Pre-tax cost of debt 5.9%, marginal. The blended book rate derived from interest expense over average total debt is 5.4%.
- Marginal tax rate 25%, the company's own stated combined rate, giving a post-tax cost of debt of 4.42%.
- Weights at market 52% equity, 48% net debt. Pro forma for EchoStar. WACC 7.0%.
Assumption block - operating and terminal
- Revenue +2.3% in 2026, the actual second-quarter rate carried forward, then 2.0% and 2.2%.
- Adjusted EBITDA +3.5% in 2026 (guidance 3% to 4%), 4.5% in 2027 and 5.0% in 2028 (guidance 5% or better).
- Adjusted EPS USD 2.30 in 2026, the guidance midpoint, then compounding at 10.5% to USD 2.86, consistent with the guided double-digit rate.
- Spectrum charge USD 5.0bn a year. The most contestable input here. Licence spend has run above USD 10bn a year since 2020 including two exceptional auctions and EchoStar, and we normalise at roughly half that. Each USD 1bn is worth USD 2.57 a share.
- Cash tax normalisation USD 1.5bn a year from 2029, as the bonus depreciation benefit unwinds.
- Terminal growth 1.0%. Below long-run US nominal GDP by design, for a mature market with a new entrant.
- Explicit period 2027E to 2031E, mid-year discounting from 1 January 2027.
| Line | Value | Line | Value |
|---|---|---|---|
| PV of 2027E to 2031E flows | 89.8 | Low: growth 0.5%, spectrum 6.5 | USD 20.00 |
| PV of terminal value | 261.6 | High: growth 1.5%, spectrum 3.5 | USD 34.74 |
| Enterprise value | 351.4 | Range midpoint | USD 27.37 |
| Less net debt, pro forma | (149.4) | Margin of safety | 25% |
| Less preferred and minorities | (22.9) | Preferred purchase price | USD 20.53 |
| Equity value | 179.1 | Current price | USD 23.71 |
| Diluted shares 2027E, bn | 6.68 | Premium to preferred entry | 15.5% |
| Base case per share | USD 26.81 | Cross-check at 7.5x 2027E EBITDA | USD 30.57 |
Our model. Preferred and minority claims of USD 22.9bn capitalise at 7% the roughly USD 1.6bn a year flowing to preferred holders and noncontrolling interests, being the gap between income from continuing operations and income attributable to common stock.
The 25% margin of safety is wider than a stable cash generator would normally warrant. It is set there for three specific reasons: pro forma leverage above three times, a spectrum reinvestment requirement the headline cash flow metric excludes, and a competitive entrant whose economics nobody can yet size. The multiple cross-check at 7.5 times, a discount to the 8.5 times peer median to reflect leverage, gives USD 30.57. We prefer the discounted cash flow, because the peer median is dragged around by two businesses in visible decline and because a multiple on EBITDA ignores the spectrum charge entirely.
| Measure | Value | Detail |
|---|---|---|
| Analysts covering | 25 | Distribution 15 buy, 9 hold, 1 sell |
| Mean price target | USD 28.71 | Our midpoint sits 4.7% below it |
| Range of targets | USD 20.00 to 36.00 | Wells Fargo at the low end, Citi at USD 34 |
| The interesting disagreement | Not the forecast | Wolfe Research upgraded to Outperform at USD 29 on 23 July 2026 on 2026 adjusted EBITDA of USD 48.2bn, almost exactly our USD 48.0bn |
Ratings and targets: Benzinga, 22 July 2026; Barchart and Investing.com, 23 July 2026; stockanalysis.com drawing on TipRanks, 6 August 2026. Tier 3 and 4. The fair value range above was constructed before consensus was retrieved. These are twelve-month targets against a three to five year holding period.
Wells Fargo initiated at Underweight with a USD 18 target on 8 July 2026 and raised it to USD 20 on 23 July. Citi held USD 34 on 24 June. The gap between them is not the operating forecast, because the bear and the bull are within a rounding error of each other on 2026 EBITDA. It is the spectrum charge and the satellite question. Set our annual charge to USD 2.5bn instead of USD 5.0bn and our midpoint moves to roughly USD 33.
Catalysts, scenarios and what the price already assumes
| Catalyst | Window | Direction | What it is worth to the thesis |
|---|---|---|---|
| Fibre passing more than 40 million locations | By 31 Dec 2026 | Positive | Confirms the build is on schedule and supports the convergence rate |
| Net leverage falling to about 3.0x | By 31 Dec 2026 | Positive | Removes the balance sheet objection; depends on the fibre subsidiary sale |
| Current USD 10bn buyback completed, next authorisation begins | By 31 Dec 2026 | Positive | About 2% a year off the share count |
| Adjusted EPS delivered inside USD 2.25 to 2.35 | FY2026 results | Neutral | First full test of the post-acquisition earnings bridge |
| Legacy segment EBITDA turning negative | After 2027 | Negative | Management's own disclosure; a cash cost until the copper estate retires |
| Copper powered down across most of the footprint | By end 2029 | Positive | Releases the fixed cost base; California approval is the stated risk to the date |
| EchoStar spectrum deployed on Ericsson 600 MHz dual-band radios | 2026 to 2027 | Positive | Turns a USD 23bn balance sheet item into network capacity |
Every row is drawn from AT&T's own statements: the releases of 28 January 2026 and 22 July 2026, and the Ericsson announcement of 5 August 2026. Tier 1.
Vertical marker = current price USD 23.71
Returns are annualised over a four-year holding period from USD 23.71 and include four years of dividends at USD 1.11. Probabilities are our judgement. The probability-weighted value is USD 26.35, 11.1% above the current price.
| Case | What has to be true | Fair value | Return a year | Probability |
|---|---|---|---|---|
| Bear | Satellite competition forces price cuts and churn above 1%; terminal growth 0.5%; spectrum charge USD 6.5bn a year | USD 20.00 | +0.8% | 30% |
| Base | Guidance delivered, leverage back to 2.6x by 2028, no material satellite share loss before 2030; terminal growth 1.0%; spectrum charge USD 5.0bn | USD 26.81 | +7.1% | 50% |
| Bull | Convergence holds churn below 0.9%, copper costs exit faster, spectrum normalises at USD 3.5bn; terminal growth 1.5% | USD 34.74 | +13.4% | 20% |
Free cash flow to the firm model, discounted at a 7.0% weighted average cost of capital with 1.0% terminal growth and a USD 5.0bn annual spectrum charge. Price is the 6 August 2026 close. The gold band is the fair value range of USD 20.00 to 34.74; the market price sits inside it, below the midpoint, and above our USD 20.53 preferred entry.
| WACC | Growth 0.0% | Growth 0.5% | Growth 1.0% | Growth 1.5% | Growth 2.0% |
|---|---|---|---|---|---|
| 6.0% | 28.59 | 32.51 | 37.21 | 42.96 | 50.14 |
| 6.5% | 24.44 | 27.69 | 31.54 | 36.15 | 41.79 |
| 7.0% | 20.88 | 23.62 | 26.81 | 30.58 | 35.11 |
| 7.5% | 17.80 | 20.13 | 22.82 | 25.95 | 29.65 |
| 8.0% | 15.11 | 17.11 | 19.40 | 22.03 | 25.11 |
Our model, spectrum charge held at USD 5.0bn a year. The base case sits at a 7.0% cost of capital and 1.0% terminal growth. Twelve of twenty-five cells clear the USD 23.71 market price.
The annual spectrum charge, worth USD 2.57 a share per billion dollars. The discount rate, where the honest range between the observed beta and our judgement moves the answer from USD 32 to USD 23. And terminal growth, which is really a proxy for whether a fourth national network exists in 2032. Everything else, including the operating forecast most of the sell side is arguing about, moves the answer by a dollar or two. A reader who disagrees with this note should aim at those three inputs and ignore the rest.
At a 7.0% cost of capital, the current pro forma enterprise value of about USD 335bn against our 2027 free cash flow to the firm of USD 20.1bn implies perpetual growth of exactly 1.0%, which is our own terminal assumption. The market and this note therefore agree on the long run. The gap between USD 23.71 and USD 26.81 is entirely the two years of cash flow growth to 2028 that the market is presently declining to pay for. That is a narrow disagreement, and it is one reason the rating is WATCH rather than BUY.
Risk register, the case against, and invalidation
| Risk | Mechanism | Effect on fair value | Leading indicator | Severity |
|---|---|---|---|---|
| Satellite entry | A fourth national carrier with a lower marginal cost of coverage breaks the three-player pricing discipline that has held since the Sprint merger | About USD 4.50 a share if 5% of postpaid service revenue reprices by 2030 | Postpaid phone churn above 1.00% | High |
| Spectrum reinvestment | The next capacity cycle requires another purchase that the company's free cash flow measure does not capture, funded by debt | USD 2.57 a share for each USD 1bn on the annual charge | FCC auction notices; secondary licence sales | High |
| Leverage and refinancing | USD 149bn of pro forma net debt carries about USD 7.6bn of annual interest against a 4.67% ten-year Treasury | About USD 0.10 of annual EPS per 100bp on USD 10bn refinanced | Rating agency outlook changes; new issue spreads | Medium |
| Copper wind-down slipping | Legacy EBITDA turns negative after 2027 and California approval could push decommissioning past 2029 | About USD 1.50 a share for a two-year delay | State regulatory rulings; Legacy segment direct costs | Medium |
| Dividend rigidity | USD 7.7bn a year is 42% of guided free cash flow but 100% of trough free cash flow | A cut would reprice the holder base, as the 2022 rebase did | Free cash flow conversion against the payout | Medium |
| Fibre subsidiary sale | The path to about 3.0x leverage this year depends on selling a controlling interest in the acquired Lumen assets to an equity partner | Roughly 0.2x of leverage and the credibility of the deleveraging plan | Any announcement of the partner or the price | Medium |
| Tax timing reversal | Cash taxes of USD 1.0bn to 1.5bn against a 25% marginal rate is a depreciation timing benefit, not a saving | About USD 3bn a year of free cash flow when it unwinds | The cash tax line in the annual cash flow statement | Medium |
The case against, argued properly
- One. The oligopoly ends. US wireless has been a good business because three players with similar cost structures stopped competing on price. SpaceX has 65 MHz of spectrum for USD 19.6bn, 12 million subscribers, USD 100bn of cash, and a model that attaches base stations to dishes already on customers' houses. It does not need to win to damage AT&T; it needs to make incumbents defend price. If 5% of postpaid service revenue reprices by 2030, about USD 2bn of EBITDA goes, worth roughly USD 4.50 a share before any multiple effect. The multiple effect would be larger, as Charter's 69% decline in twelve months demonstrates.
- Two. The spectrum treadmill is permanent and the cash flow metric hides it. Since 2020 AT&T has spent more than USD 60bn on licences and has now added USD 23bn more, while reporting free cash flow that excludes every dollar of it. In 2026 it will report about USD 18bn of free cash flow and increase net debt by roughly USD 30bn. At a USD 6.5bn normalised charge rather than USD 5.0bn, fair value falls to USD 20.00.
- Three. The balance sheet has no slack at a trough. At USD 40bn of adjusted EBITDA, free cash flow of about USD 7.6bn exactly covers the USD 7.7bn dividend and nothing else. The buyback stops, deleveraging stops, and the next spectrum purchase is entirely debt funded at 3.7 times leverage. AT&T does not break, but it loses every option it has, and a company with no options trades at a lower multiple than one with several.
The case for, argued properly
- One. Convergence is a genuine structural advantage and it is compounding. 42.5% of advanced home internet households take AT&T wireless, and converged households churn less. Only AT&T has both a national wireless network and fibre at this scale; Verizon's fibre footprint is smaller and T-Mobile's is nascent. If the rate reaches 50% by 2028 the churn benefit alone supports the guided EBITDA path. Plausibility high, because the mechanism is already visible in the churn number.
- Two. The copper estate is a cost pile waiting to be removed. Legacy revenue fell 25.9% year on year while its direct costs fell only 10.8%, which is why segment EBITDA turns negative after 2027. Once the network retires, a fixed cost base disappears permanently from a company that no longer needs it. Plausibility medium, since the timetable depends on regulators not aligned with AT&T. Worth roughly USD 1.50 a share if it lands on time rather than two years late.
- Three. The share count finally works for holders. Roughly USD 24bn of buybacks across 2026 to 2028 on a USD 162.5bn market value removes about 11% of the shares, worth about USD 2.60 a share at a constant multiple, and the largest contributor to the guided double-digit adjusted earnings per share compound rate. Plausibility high: the authorisations are in place and the 2026 pace has been raised to USD 10bn.
Invalidation: what would tell a holder they are wrong
- The thesis is wrong on the downside if postpaid phone churn exceeds 1.05% for two consecutive quarters, adjusted EBITDA growth falls below 2% in a full year through 2028, net leverage exceeds 3.3 times at any year end, the dividend is cut, or a further spectrum purchase above USD 10bn is announced without a matching disposal.
- It is wrong on the upside, and the WATCH should be closed rather than converted, if the price passes USD 30 without a lower spectrum charge or a resolution of the satellite question.
- The rating moves to BUY at USD 20.50 or below, being the fair value midpoint of USD 27.37 less a 25% margin of safety.
Portfolio fit, entry strategy and the decision
| Property | Measured | Constraint it imposes |
|---|---|---|
| Liquidity | About USD 1.81bn traded a day, free float 99.9% of shares outstanding | Not a constraint at any realistic private position size |
| Concentration | More than 100 million customer relationships, no client of consequence | Not a constraint |
| Volatility | 52-week range USD 19.89 to 29.79, a 50% spread, against a five-year beta of 0.42 | The beta understates the realised risk. Size on the range, not the beta |
| Correlation and overlap | A top-fifty S&P 500 constituent, inside every broad US index fund, value tilt and dividend strategy | A holder with a large US core is adding concentration, not diversification |
| Drawdown | −15.6% modelled from USD 23.71 to the bear case; the 52-week low is 16.1% below today | The tolerance a holder needs before buying, not after |
Volume, float and range from stockanalysis.com drawing on S&P Global Market Intelligence, 6 August 2026. Taken together these are the properties of a 2% to 4% band of a diversified equity portfolio. That is a statement about the security, not advice to any reader.
AT&T is a US-situs security paying a US-source dividend. The headline 4.68% yield is gross, and the net figure after US dividend withholding is materially lower for a holder in a jurisdiction with no US tax treaty; the shares are also a US-situs asset for US estate tax purposes. Both should be confirmed against current rates before sizing, and both argue for taking equivalent exposure through an Irish-domiciled vehicle where one exists. On a single name there is none, which is itself a reason to keep the band at the lower end.
Entry strategy and monitoring
- No entry at USD 23.71, which sits 15.5% above the preferred purchase price, though 13.4% below the fair value midpoint.
- Preferred purchase price USD 20.53, the midpoint of USD 27.37 less a 25% margin of safety. At that price the expected annualised return over four years is 12.9%: 7.5% capital return to the midpoint plus a 5.4% yield on cost. At today's USD 23.71 the same arithmetic gives 8.3% against a 4.67% risk-free rate, a premium of 3.6 points for owning a levered telecom facing a new entrant. That is the whole case for waiting.
- Build in two tranches: half at or below USD 20.50, and half on either a confirmed net leverage print at or below 3.0 times at the full-year results or a quarter with postpaid phone churn at or below 0.90%.
- Monitor quarterly: postpaid phone churn, the convergence rate, net leverage, and the spectrum line inside investing activities, which is where the cost this note is most concerned with actually appears. Annually: the cash tax line, Legacy segment EBITDA, and copper decommissioning approvals. Continuously: any SpaceX announcement on Starlink Mobile pricing, coverage or a carrier agreement, since a commercial launch with published pricing turns an unquantifiable risk into a quantifiable one.
Decision checklist - passes
- Business understandable: three products, one recurring revenue model - Pass
- Moat identified with a mechanism: narrow, licence and sunk network - Pass
- Balance sheet survives a downturn: 5.3x interest cover at trough - Pass
- Management trustworthy on operations: every 2025 target met or beaten - Pass
- Risks tolerable and monitorable: quantified, with named indicators - Pass
- Position sizeable without market impact: USD 1.81bn traded a day - Pass
Decision checklist - failing
- Earnings quality clean: free cash flow excludes spectrum, an 11.7% effective tax rate against a 25% marginal rate, and a third of 2025 earnings per share from a disposal - Fail
- Price below fair value less the margin of safety: trades 15.5% above the USD 20.53 preferred entry - Fail
WATCH, trigger USD 20.50. Six of eight tests pass. The two that fail are earnings quality and price, and only one of those can be fixed by waiting. A reader whose required return for a large, liquid, dividend-paying telecom is 6.0% rather than 7.0%, or who believes the normalised spectrum charge is USD 2.5bn rather than USD 5.0bn, would call this cheap here. That reader is not making an error, they are using a different input. Both are printed in the valuation assumption block above. Argue with those.
Appendix
| Rating | Definition |
|---|---|
| BUY | Price at or below the fair value midpoint less the stated margin of safety, with business, balance sheet, governance and valuation tests all passing |
| WATCH | Quality tests pass but the price does not, or a catalyst must be confirmed first. A WATCH must name its trigger |
| AVOID | Fails on business quality, balance sheet, governance or valuation, with no realistic path to the required return |
The equivalent house convention is BUY, HOLD, SELL. WATCH replaces HOLD because these notes inform a decision to buy rather than advise an existing position. AT&T is rated WATCH, trigger USD 20.50.
| Source | Tier | Used for | As of |
|---|---|---|---|
| AT&T second-quarter 2026 earnings release and segment tables | 1 | All Q2 2026 figures: revenue, EBITDA, segments, subscribers, churn, debt, cash flow, guidance | 22 Jul 2026 |
| AT&T fourth-quarter and full-year 2025 earnings release and call | 1 | FY2025 income statement, cash flow, leverage, ARPU, 2026 to 2028 guidance | 28 Jan 2026 |
| AT&T fourth-quarter and full-year 2024 earnings release | 1 | FY2024 revenue, adjusted EBITDA, free cash flow, net debt, goodwill impairment | 27 Jan 2025 |
| AT&T Analyst and Investor Day release, Form 8-K exhibit | 1 | The December 2024 target set used in the governance section | 3 Dec 2024 |
| AT&T press release, EchoStar spectrum acquisition closing | 1 | USD 23bn consideration, 30 MHz of 3.45 GHz and 20 MHz of 600 MHz, closing date | 28 Jul 2026 |
| AT&T and Ericsson, 600 MHz dual-band radio announcement | 1 | Deployment path for the acquired spectrum | 5 Aug 2026 |
| SpaceX second-quarter 2026 earnings call remarks, via Stocktwits and TipRanks | 3 | The USD 600bn industry billing figure, the 65 MHz terrestrial network plan, USD 100bn cash, 12m subscribers | 4 Aug 2026 |
| Reuters | 3 | Independent confirmation of the EchoStar closing | 28 Jul 2026 |
| Trading Economics and CNBC | 3 | US 10-year Treasury yield of 4.67% | 6 Aug 2026 |
| Benzinga; Barchart; Investing.com | 3 | Analyst ratings, targets and the Wolfe Research 2026 estimates | 22 to 23 Jul 2026 |
| Moody's Baa2 affirmation on the EchoStar announcement, via financial press | 3 | Credit rating and the agency's stated leverage expectation | 27 Aug 2025 |
| Morningstar | 4 | Company profile and the more than USD 60bn of licence spend since 2020 | Aug 2026 |
| stockanalysis.com, via S&P Global Market Intelligence and TipRanks | 4 | Price, ratios, 52-week range, beta, share count, revenue series, peer data, consensus | 6 to 7 Aug 2026 |
| Yahoo Finance | 4 | Form 13F institutional holdings | 30 Sep and 31 Dec 2025 |
| GuruFocus | 4 | AT&T five-year median price to earnings of 9.0x | 7 Jul 2026 |
What desk research cannot answer
- The executive compensation structure. We could not confirm which metrics the incentive plan pays on, and with insider ownership at 0.11% that plan is the only alignment mechanism there is.
- The audit and related party disclosures. The auditor's report, any emphasis-of-matter paragraph and the related party note were not reviewed, so the audit-flags block was dropped rather than guessed.
- The debt maturity schedule by year. Not reproduced here; the blended cost of debt was derived from interest expense over average total debt rather than read from the note.
- The size of the Starlink Mobile threat. Nobody outside SpaceX can currently size it, because there is no published pricing, launch date or coverage commitment. That is where the remaining edge on this name sits, and it will be resolved by disclosure rather than by analysis.
Basis of analysis. Financial figures are drawn from AT&T's earnings releases of 27 January 2025, 28 January 2026 and 22 July 2026, and the Analyst and Investor Day materials of 3 December 2024. Market data, ratios and peer comparatives are as of the close on 6 August 2026 unless a different date is stated in the source line. All figures are in United States dollars and no currency conversion was applied. Forecast columns are marked E and every estimate traces to the assumption block in section 06. The fair value range was built before analyst consensus was retrieved, and blocks that could not be verified from a named source were removed rather than estimated. The analyst holds no position in AT&T Inc.
Analyst certification. The views expressed accurately reflect the analyst's own assessment of the securities and issuer discussed, formed from the sources above on the date shown. No part of any compensation was, is, or will be related to the specific recommendation or views expressed. This note was prepared independently and no compensation was received from the issuer or any related party.
This note on AT&T Inc. was published on 7 August 2026 and reflects public information available to that date. Company results, market prices and analyst estimates change; readers should verify current figures independently before making any decision.
This document is independent research prepared for informational purposes. It is not investment advice, not a recommendation to buy or sell any security, and does not constitute an offer or solicitation. Figures are drawn from public sources believed reliable as of the date shown and may change without notice. Anyone acting on this material does so at their own risk and should seek their own professional advice.