Ally Financial Inc. ALLY

Key data

The business

Ally is the direct-bank descendant of GMAC, the old financing arm of General Motors. It has no branches. It gathers deposits from savers online, then lends most of that money to people buying cars, arranged through dealers across the country. The difference between what borrowers pay and what savers earn is the engine. Around that sit three smaller pieces: an insurance arm that covers cars and dealers, a corporate-finance unit that lends to mid-size companies, and a run-off mortgage book.

The moat is the deposit franchise. Ally was the direct bank before direct banks were normal, and money that arrives by app, with no branch to walk into, is cheap and unusually sticky. The thing a customer holds is not a passbook but a phone screen showing a savings rate a few tenths above the branch banks. That funding cost is the whole game: cheap deposits are what let a lender survive the quarter its borrowers stop paying.

The numbers

The last five reported quarters, with the credit cost sitting alongside the top line. Q4 2025 is derived as the full year less the filed nine months.

Net revenue in Q2 grew about 10% over the year-ago quarter and net income 17%. On the company's own adjusted measure the quarter was $1.21 against a $1.22 estimate, a hair light, after three straight beats, including a large one in Q1. Read the sequence and the story is a business that has stopped shrinking and started widening its margin, one quarter at a time. But the credit line runs the other way: the amount set aside for loans going bad was $897 million in the first half against $575 million a year earlier, up 56%. The cost-of-credit creep the last look flagged has not reversed in the weeks since; it is still the live tension in the print.

Here is the whole argument. Earnings per share fell from $8.22 in 2021, when the used-car boom and near-zero funding costs made every auto lender look brilliant, to $1.80 in 2024, as that boom faded and Ally paid up for deposits. 2025's $2.37 was the first turn up, and even that was held down by a $305 million goodwill write-off and a $438 million loss on investments in the first half, much of it from selling securities. Consensus has 2026 earnings at $5.23 and 2027 at $6.42. That is not a growth fantasy, it is the reported number catching up to earning power once the one-time hits stop repeating and the deposit repricing does its work. What this memo believes that the tape does not: normalized earnings are near $5 already, so 0.9x book and a single-digit forward multiple are pricing a lender that never mends. The print that settles it is whether 2026 GAAP net income clears roughly $1.5 billion. The half-year run-rate says it is close.

Management

Michael Rhodes is CEO, with insider data showing an open-market purchase of about $1.0 million of stock in January 2026, near the low end of the year's range. Other insiders were net sellers over the last twelve months, $3.4 million against $1.5 million bought, led by a $1.8 million sale in April, with plan status not disclosed on the feed. Capital return tells the same recovery story as the earnings: buybacks ran near $2.0 billion in 2021, then went dark from 2023 through 2025, dropping to $33 to $59 million a year while the bank rebuilt its cushion. In the first half of 2026 repurchases restarted at $147 million against $34 million a year earlier. Management is spending again because it believes the capital is there.

How it fails or surprises you

Auto credit turns. Provisioning is up 56% in the first half, and Ally's borrowers skew toward the part of the car market that breaks first. If unemployment rises or used-car values fall, charge-offs on the auto book overrun the reserve and eat the earnings recovery whole. The tell is provision as a share of net revenue, and the net charge-off rate in the next 10-Q, climbing two quarters running.

Deposits reprice down faster than loans (right tail). Ally's funding cost moves with short rates, its loan yields with slower auto pricing. If deposit costs fall while the back book keeps rolling into higher-yield originations, net interest margin widens and net income steps up before consensus catches it. The market pays 0.9x book and is not paying for this. It shows first in the margin and in sequential net income.

The GAAP-to-core gap closes, or it doesn't. Reported 2025 EPS was $2.37, consensus 2026 is $5.23, and the whole bridge is one-time items not recurring. If a fresh write-down appears or margin stalls, the "one-timers" look like a pattern and the re-rating never comes. Watch whether 2026 GAAP actually lands near $5, not the adjusted figure.

Closing thoughts

Whether 2026 GAAP earnings reach $5 will settle whether the recovery is real. The market is treating Ally as a broken auto lender at nine-tenths of paper worth while its own numbers say the margin is mending and the write-offs of 2025 were events, not a trend. The print that converts the debate is 2026 GAAP earnings against the roughly $5 the Street models. Land there and 0.9x book on a normalizing bank is too cheap. Miss it because credit ran away and the discount was correct. The fatter tail leans up from here, because the downside is largely visible in the provision line already while the margin recovery is not yet in the price, but what is at risk if credit turns is real: a lender's worst quarter can erase a year of the recovery.

The bet is still that Ally makes car loans that earn more than its online deposits cost, and that the gap widens as the expensive deposits roll off. What breaks it is the borrower who stops paying before that gap widens. The one pair of numbers that tells you which way it broke: provision for credit losses against net revenue, quarter after quarter. If provisioning keeps outrunning the top line into 2027, the recovery was a mirage.