AT&T INC. (T) stock intrinsic value fair value estimate & owner earnings analysis

AT&T is now a focused U.S. connectivity company: wireless service to about 120 million subscribers, a growing fiber-to-the-home and fixed-wireless (Internet Air) broadband business, and business networking, plus wireless in Mexico. Its ten-year record before 2022 describes a different company. It bought DIRECTV (2015) and Time Warner (2018) for roughly $128 billion combined and then unwound both: WarnerMedia was spun off in April 2022 and the DIRECTV stake was sold in 2025, with the dividend cut nearly in half in 2022. Only 2022 onward is comparable with the business as it stands, and that is the window used. How we measure it. We start from operating cash flow, remove swings in receivables, payables and inventory, deduct stock compensation, and deduct all capital spending, including the multi-year vendor-financed equipment payments that sit outside the reported capital spending line ($1.2 to $5.7 billion a year). Growth capital is separated from maintenance capital only to the small extent revenue has grown. We then add back cash interest, so the stream shows what the business earns before its lenders are paid, and subtract the debt afterwards. On that basis the business produced about $19.8 billion in 2022, $22.7 billion in 2023, $24.3 billion in 2024 and $26.0 billion in 2025. We anchor on the middle of those four years, about $23.5 billion a year, rather than the latest and highest year: 2022 reflects the peak of the 5G build, 2025 reflects a year of low vendor-financing payments, and 2026 capital investment is guided up to $23 to $24 billion from $22.0 billion. First-half 2026 data agree with the lower anchor: operating cash flow was $18.4 billion against $18.8 billion a year earlier, and capital spending rose to $10.6 billion from $9.2 billion. The latest full-year growth in this stream was about 7%; the base case assumes none in year one, because the 2026 capital step-up is already visible. Spectrum. AT&T has spent about $62 billion on wireless licences since 2021 (C-band $25 billion in 2021 and $10 billion in 2022, plus the EchoStar purchase of 600 MHz and 3.45 GHz licences for about $23 billion, which closed July 28, 2026). Licences do not wear out, but capacity needs recur. We treat future spectrum as a continuing cost beginning in year two: $5 billion a year in the base case, $6 billion in the bear case (spending is cut under pressure) and $2 billion in the bull case. The EchoStar price is already counted in the debt below. What the company owes. Debt at June 30, 2026 was $144.0 billion ($134.6 billion long-term and $9.3 billion due within a year), and the EchoStar closing adds roughly $23 billion funded partly by a term loan and partly from cash. Outside investors hold $16.0 billion of preferred interests in AT&T subsidiaries, which we treat as a claim ahead of the shareholders. Against that we credit about $14.6 billion of cash beyond operating needs, and about $18.9 billion because the debt carries an average 4.2% interest rate, locked in for years: $9.0 billion matures in 2026 at about 3.1%, roughly $7 billion a year in 2028 through 2030, and $106 billion beyond that. Net, the business owes about $149.5 billion, or $21.5 per share on 6.95 billion diluted shares. Operating leases, pension contributions and payables are already charged inside cash flow and are not subtracted again. Base case: business worth about $340 billion ($48.9 a share), less $149.5 billion owed, leaving $27.4 a share. Cash flow stays at the anchor in year one, grows 1.3% a year for four years (the rate at which revenue actually grew from 2023 to 2025: $122.4 billion to $125.6 billion), then the growth fades to nothing by year ten. After year ten we assume flat cash flow after spectrum, because the customer base and network are a long-standing structural feature and nothing more can be named. We left out subscriber growth from the Lumen fiber business bought on February 2, 2026 (a plan to sell part of it to a partner is not complete), share gains, and any benefit from share repurchases, because none is secured or proven at AT&T. Bear case: business worth about $170 billion, less $149.5 billion owed, leaving $3.0 a share (about $20.5 billion of equity in total). The story is a repeat of 2015 and 2016, when AT&T Mobility service revenue fell 5.5% in each year as customers moved to cheaper plans. With network costs largely fixed, we cut cash flow before interest by 12% in year one and 25% by year two, hold it there, spend $6 billion a year on spectrum from year two and let it decline 2% a year after year ten. Cash flow after spectrum is still about $11 billion against roughly $6.6 billion of annual cash interest, so the debt is serviceable, but almost nothing is left for shareholders. Bull case: business worth about $430 billion, less $149.5 billion owed, leaving $40.4 a share. Cash flow starts at $24.5 billion after the 2026 capital step-up and grows 2.5% a year for four years as the fiber footprint is filled in (Internet Air reached 1.5 million connections after adding 875,000 in 2025, and Lumen's roughly one million fiber customers are now owned), then tapers to flat; spectrum spending falls to $2 billion a year. What changed since the last report. The earlier figures ($8, $23 and $46) were prepared on a different basis and are not comparable. In the business itself, the EchoStar purchase closed and added about $23 billion of debt, the Lumen fiber purchase closed, and about $4.7 billion of shares were repurchased in the first half of 2026, taking diluted shares from 7.18 billion for 2025 to 6.95 billion in the second quarter. Net debt is higher; the cash flow anchor is lower than 2025 reported. What would make this wrong. Each 10% error in the cash-flow level moves the base value by roughly $5 a share; spectrum purchases averaging $8 billion a year rather than $5 billion would remove about $7 a share. The two together would put the base value near $16. The 2015-2016 style wireless price reset is the main risk the filings point to, and the leverage makes the shares move about twice as much as the business does. Margin-of-safety price. We are fairly sure of the level of cash flow: it has been between $19.8 and $26.0 billion for four years, and a 5G-build capital cycle is the main swing. The worst plausible outcome is a repeat of the 2015-2016 pricing reset coming on top of an annual spectrum bill that must still be paid; the business would survive (cash flow stays above interest), but the shares would be worth very little, and the $16 billion of subsidiary preferred interests ranks ahead of shareholders. If we are wrong on the cash-flow level by 10% and on spectrum, the shares would be worth about $16, and in that case the price still implies a double-digit annual return over about four years, given the $1.11 annual dividend, which is covered more than twice by cash flow after all capital spending, and continuing share repurchases. The margin-of-safety price is $13: at that price the dividend alone is 8.5% a year, the price implies about 17% a year over four years if the cash flow is 10% lower and spectrum costs are higher than assumed, and it reflects a wide range of outcomes between $3 and $40.