AT&T is buying back $10 billion of its own stock at 7x earnings. Is the seller making a mistake?

The quarter to September 2026 had a $0.60 bar and a $4.9 billion cash test

AT&T ($T) reported the quarter to September 2026 on 21 October 2026, before the market opened. The bar was $0.60 of adjusted earnings per share on $31.69 billion of revenue, against $0.54 and $30.71 billion in the same quarter of 2025, and options traders had priced a 4.1% move either way. The shares traded at $23.45 by the close, up 1.08% - a muted reaction next to the 3.5% jump after the July 2026 release and the 4.7% jump after the January 2026 one.

Three things decide whether the quarter was good. Free cash flow had to hold near the $4.9 billion of the quarter to September 2025, the level management guided as roughly stable on the 22 July call, with the strong growth pushed into the fourth quarter. Customer adds had to clear 405,000 postpaid phones and 288,000 fiber connections from the comparable 2025 quarter. And the full-year plan of $18 billion+ free cash flow and $2.25-2.35 adjusted EPS had to survive intact. The record going in was lopsided: EPS beat the estimate in three of the four prior quarters, by 3.6% to 13.0%, while revenue missed in two of them.

Three deals in 100 days: $23 billion out, $2 billion in, and a fiber venture with no price tag

On 28 July 2026 AT&T paid about $23 billion in cash for roughly 50 MHz of EchoStar spectrum. At 30 June 2026 net debt was $126.4 billion on total debt of $144.0 billion, or 2.68x adjusted EBITDA; the 22 July call put the post-close ratio at around 3.2x, with a return to about 2.5x within roughly three years. In plain terms: the company now owes a little over three years of its operating profit, up from under two and three-quarters.

Two deals since then pull the other way. On 13 October 2026 AT&T agreed to sell AT&T Mexico to Grupo Televisa and Fintech Holdings for $2 billion in cash, closing in the first quarter of 2027. The unit produced $1.22 billion of revenue and $227 million of EBITDA in the quarter to June 2026, so the price is roughly 2.2x annualised EBITDA - cheap, but it removes a low-margin business from a 3.2x balance sheet. On 6 October a 50/50 fiber joint venture with Global Infrastructure Partners and CPP Investments took in the Gigapower and Forged Fiber 37 assets, the former Lumen footprint of about 4.5 million locations, roughly 25% penetrated against 40% in AT&T's own territory; terms were not disclosed and closing is set for the first half of 2027. The step to 3.2x leverage is the balance-sheet reason the Street sits on hold while the cash flows argue otherwise.

Fair value $30.69: an 11.1% cash yield at 7.4x earnings is what lifts the DCF

The FPI fair value is $30.69, 30.9% above the $23.45 price. The relative side is simple: AT&T trades at 7.4x earnings against a telecom industry median of 15.4x, at 5.7x EV/EBITDA against 6.5x, and with a free cash flow yield of 11.10% against 9.08% - for every $100 of shares the company generates $11.10 of cash after capital spending. At the same time its 16.9% net margin and 19.5% return on equity beat the medians of 9.1% and 12.6%. A stock priced below its peers on every multiple while out-earning them is what a peer comparison marks up.

The cash side does the heavier lifting. The company's own plan is free cash flow of $18 billion+ in 2026, rising to $19 billion+ in 2027 and $21 billion+ in 2028, with $45 billion+ of dividends and buybacks over 2026-2028 - about 28% of the $159 billion market cap handed back in three years. Those are the kind of inputs that lift a DCF. The drag is real too: return on invested capital of 6.0% sits below the 6.2% median and three-year revenue growth is 1.3% a year. The multiple has compressed from 15.0x at the end of 2024 to 8.1x at the end of 2025 and 7.4x now, while the price went from $22.77 to $24.84 to $23.45 - earnings grew, the price did not follow.

Street vs model: $27 median, $24 from Bernstein, $36 from KeyBanc, $30.69 from FPI

The consensus on 21 October 2026 is hold: 1 strong buy, 26 buy, 29 hold, 6 sell. Targets run from $20 to $36, with a median of $27.00 and a mean of $27.50; the five targets of the last month average $28.80. FPI's $30.69 is 11.6% above the mean and 13.7% above the median. The mean itself is 17.3% above the price, so the hold rating is about confidence, not arithmetic.

The latest moves point in opposite directions. On 13 October 2026 Bernstein cut AT&T to Market Perform and its target to $24 from $25, citing SpaceX's push into broadband and wireless and fourth-quarter seasonality, with the stock at $22.49. KeyBanc reaffirmed Overweight and $36 on 7 October. Scotiabank trimmed to $27.50 on 2 October and $26.50 on 9 October. On 21 September HSBC went to Reduce and BNP Paribas started at Outperform with $30 - the same day, the same facts, a $12 spread in targets.

The gap has two causes. The Street applies a satellite-competition discount and a 3.2x leverage penalty; the model weights an 11.1% cash yield and a 4.78% dividend. The number that settles it is churn: 0.86% of postpaid phone customers left per month in the quarter to June 2026, down from 0.92% a year earlier. If Starlink were pulling customers, churn would show it first.

Three numbers to watch: $25.09, 0.9% churn and $18 billion

A re-rating up needs the 2026 free cash flow to land at $18 billion+ with the promised fourth-quarter growth, churn held below 0.9%, buybacks continuing at the roughly $2.2 billion-a-quarter pace of the quarter to June 2026, and the price recovering the 200-day average of $25.09 - a 7% move from $23.45. A re-rating down comes from churn above the 0.92% of the quarter to September 2025, free cash flow guidance slipping below $18 billion, or net debt/EBITDA still above 3x going into 2028.

Dated checkpoints: the $0.2775 quarterly dividend due on 2 November 2026, the Mexico closing in the first quarter of 2027 and the fiber venture closing in the first half of 2027. With a beta of 0.43 the stock moves less than half as much as the market, and it sits 20.9% below its 52-week high of $29.44. Whoever sells at $23.45 is betting that Starlink and 3.2x leverage outweigh an 11% cash yield. The company has $10 billion on the other side of that trade.