Atlanticus (ATLC) Q3 2026 earnings review
Atlanticus profit doubles, but its original lending business slows
Atlanticus, which funds credit cards for Americans with weaker credit, earned far more while its original business slowed. Excluding one-offs from two asset sales, profit per share reached $3.00, about double a year ago. Mercury, the card business it bought last year, stopped shrinking. But by our math, loans outside Mercury grew about 18%, down from 26% a quarter earlier.
| Sales growth excluding the Mercury purchase | +18% +60% including Mercury |
|---|---|
| Profit per share, excluding one-offs | $3.00 $1.48 a year ago |
| Share of loans written off, yearly pace | 17.3% 17.7% last quarter |
| Return-on-equity target | Unchanged: 20% or more 30.1% this quarter, excluding one-offs |
⚖️ Verdict: 🟢 Bullish
The story got better, because more profit now comes from lending than from accounting estimates. By our math, estimate gains made up about 24% of pretax profit, against about 71% last quarter. The bad news: card customers fell behind on payments more often, as Atlanticus reaches for riskier borrowers.
The question now is whether growth outside Mercury levels off in the high teens or keeps fading. If it levels off, the bigger loan book keeps lifting profit; if it fades, Atlanticus must take on more risk to grow. Year-end loan figures, due early next year, will tell.
🐂 Bull Case
Mercury Card Customers Now Pay Higher Rates
Mercury is the card business Atlanticus bought in September 2025, with about $3 billion of loans to near-prime customers, those just below the best credit grades. Since then, Atlanticus has made what it calls "product, policy and pricing changes" to earn more on each dollar lent.
"The Mercury acquisition continues to deliver better-than-modeled performance on both the back book and new origination volumes," chief executive Jeff Howard said. This quarter supports him on rates and on size, because the Mercury book stopped shrinking.
- Average rate on general-purpose cards: 30.4%, up from 28.7% a year ago
- Yearly income on all managed loans: 34.9% of the balance, against 34.2% in June
- Mercury loans: $3.1 billion, against $3.05 billion in June
Mercury holds over 40% of the loans Atlanticus manages, so higher rates there lift income across the group. But some changes "will take several quarters to be fully realized", the company says, so more of the gain lies ahead.
What to watch: the yearly income rate on managed loans in the fourth-quarter 10-Q. A rate that holds near 35% while the Mercury book stays above $3 billion would show customers accept the new prices.
Underlying Profit Doubled as Borrowing Got Cheaper
Excluding one-offs from selling its car-loan unit and the Mercury name, Atlanticus earned $57.0 million for shareholders. That is 104% above a year earlier, so profit grew far faster than sales.
The largest cost explains most of it. Atlanticus books loan write-offs and changes in what its loans are worth as one line, and that line took a smaller share of revenue this time.
- Write-offs and loan value changes: 53.9% of revenue, down 3.2 percentage points from a year ago
- Interest on debt: 15.2% of revenue, down 0.3 points
- Operating costs: 20.9% of revenue, down 0.3 points
Cheaper funding helped as well. Atlanticus repaid three credit lines this spring, one of them fixed at 9.51%. Its interest bill then eased to 6.7% of average loans, from 7.2% in June. The other side of the case still holds. Debt stood at $7.95 for every dollar of equity, against $8.94 in June, so the lender still depends on open debt markets.
What to watch: the interest bill as a share of average loans in the fourth quarter. A ratio that stays near 6.7% would show the cheaper funding lasts as Atlanticus borrows to grow.
Online Marketing Brought a Record Number of Customers
Atlanticus now serves 6.7 million accounts, 18% more than a year ago. Chief executive Jeff Howard said its bank partners opened a record number of new accounts for the third quarter in a row.
Howard credits online marketing. Atlanticus builds separate models for each digital channel and moves money to the ones that work, he said, while mail offers face "heightened competition on the direct mail channel".
This matters because new accounts become next year's loans and fees. General-purpose card accounts reached 5.0 million, from 4.6 million in June. Card purchases that Atlanticus funded rose 7.6% from the second quarter.
What to watch: general-purpose card balances in the fourth quarter. If accounts keep climbing while balances stay near $5.1 billion, the new customers borrow less than the old ones.
🐻 Bear Case
Card Customers Fell Behind on Payments More Often
Atlanticus funds credit for Americans with weaker credit, so its main risk is that more of them stop paying. "Credit remained stable quarter-on-quarter," finance chief Bill McCamey said. The yearly write-off rate backs him: it eased to 17.3% of average loans from 17.7% in June.
Early late payments tell a different story. On general-purpose cards, loans a month or two overdue reached 6.6% of the book, against 5.8% both in June and a year earlier. Such loans tend to become write-offs within months.
Chief executive Jeff Howard said the rise in new accounts is "not the result of risk expansion". On the call, though, he described a "broader range of consumers" who "tend to have, on average, higher delinquencies and higher yields". Atlanticus itself expects late payments to "marginally increase".
That makes the rise partly a deliberate trade of risk for income. Howard said customers remain "resilient and rational" despite inflation and gas prices, but the card numbers moved first.
What to watch: the share of card loans a month or two overdue in the fourth-quarter 10-Q. A reading above 6.6% would show the new customers paying worse than their higher rates allow for.
Growth Outside Mercury Slowed for a Third Quarter
Atlanticus's own loan book, built before the Mercury purchase, is where its long-term growth must come from. By our math, those loans grew about 18% from a year ago. That is down from 26% in June and 35% in March.
Store cards, which retailers offer at checkout, slowed most. The 10-Q says "some retail partner programs moderated in the third quarter of 2026", and one partner still dominates.
- Store-card loans bought in the quarter: $385.6 million, down 11.7%
- From the largest retail partner: $278.3 million, down 17.5%
- Top five partners' share of store-card loans: 86.8%
Each new customer also costs more. Chief executive Jeff Howard expects costs per new account "will continue to go up", and marketing spending rose 19% from June.
What to watch: loans outside Mercury at year-end, which needs a new Mercury balance figure. Growth below the high teens would mean Atlanticus pays more per customer for a slower result.
Profit Still Includes Gains From Accounting Estimates
Atlanticus values its loans at what it expects to collect, so a change in those estimates moves profit at once. Last quarter, estimate gains and a cut in the payout owed to Mercury's former owners made up most of pretax profit.
This quarter they mattered far less. Estimate changes added $10.6 million and the smaller payout another $8.0 million. By our math, that is about 24% of pretax profit excluding the Mercury name sale, down from about 71%.
The risk remains, because the same model turns gains into losses if customers pay worse. More card loans running late is the kind of change that would push estimates the other way.
What to watch: the estimate-change line in the fourth-quarter 10-Q. A loss there would show the valuation model cutting profit instead of adding to it.
Risks this quarter didn't answer
Five standing risks got no new figures in this quarter's results, so each one still waits for a specific disclosure from Atlanticus or its regulators.
- Regulation: no update on rules for bank-partner lending or card fees; new federal or state action would move it.
- Mercury systems: the switch to Atlanticus technology runs into early 2027; a completion date would close it.
- Mercury profit: Atlanticus gives revenue but no separate profit figure, nor progress on its 2027 per-share target; a number would test both.
- Hanna family interests: preferred shares held by family trusts can convert into common stock; any change in terms would move it.
- Excluded store-card loans: loans whose losses retailers cover stay out of late-payment rates, unsized; a figure would show their weight.
👓 Other Themes
Atlanticus Sold Car Loans and the Mercury Name
Atlanticus sold CAR, its car-loan unit, for about $71.2 million and booked a $20.1 million pretax gain. It also sold the Mercury brand name for $27.5 million but kept the card loans. It may use the name to sign up customers for one more year, which Howard called "ample time for the necessary rebranding". Atlanticus has not said what renaming the card will cost.
💲 Other KPIs
The ratio rose for a third quarter. It compares what Atlanticus says its loans are worth with what customers owe, so a higher figure means it expects to collect more. It stood at 95.6% at the end of 2025 and 96.2% a year ago.
Almost unchanged, up 0.1% from a year ago. Atlanticus sold 152,132 new shares for $14.2 million in the quarter. It bought back only 4,402, so repurchases stayed near zero for a third quarter. The count includes about 4.0 million shares from convertible preferred stock.
Cash climbed 14% from June and 40% from a year ago, helped by the car-unit sale. The lender needs part of it soon, because $115.5 million of notes fall due on November 30, 2026.
🔮 Guidance
Unchanged. The plan still has card loans rising each quarter and faster than store-card loans. Card balances grew 3.2% from June. So far this year, though, they are up only 0.9%, against 7.0% for store cards, so the fourth quarter must close the gap.
Unchanged. Atlanticus still plans only a small rise for the year, because purchases tied to one large retailer are slowing. These balances usually fall back outside the second and third quarters, the company said, so part of this year's rise may fade.
Unchanged. Atlanticus still expects its interest bill to climb as it borrows to fund more loans. The bill reached $118.0 million, 57% above a year ago, but it came in below the second quarter after the refinancing.
Unchanged. The plan has marketing spending grow this year as the company pays more to win each customer. Spending reached $57.5 million, 63% above the same quarter of 2025, and timing depends on the economy, the company said.
Unchanged. Pay costs stay above last year because Mercury brought its own staff. They reached $25.6 million, 52% above the third quarter of 2025, though the bill came in below the second quarter.
Unchanged. Atlanticus still expects new prices and rules on Mercury cards to lift revenue. Mercury added $205.3 million of revenue from a year ago, but Atlanticus owned it for under three weeks of that earlier quarter, so the gap overstates the effect.
Unchanged. Atlanticus aims to earn at least 20 cents a year on each dollar of shareholder money. Excluding one-offs it earned 30.1% this quarter, the fourth quarter in a row above the target. If the pattern holds, the full year clears the target comfortably.
Unchanged. Atlanticus expects late payments to rise slightly as bank partners offer cards to a wider range of customers. It also expects the write-off rate to improve slightly for the rest of 2026, then return to "historically normalized levels" in 2027.
❓ Key Questions
What will renaming the Mercury card cost?
Atlanticus sold the Mercury name and may use it for new customers for only 12 months. Renaming a large card program can raise marketing costs and lose customers, and the company gave no budget or timetable.
Why did the largest retail partner buy less again?
Purchases from the biggest store-card partner fell from a year ago for a second quarter. Five partners hold most store-card loans, so a change in that one program's terms would move the whole channel.
How much does Mercury earn on its own?
Atlanticus reports Mercury's revenue but not its profit. Without that figure, readers cannot tell whether the new prices pay off, or test the per-share gain the company targets from the deal in 2027.
Why did the expected payout to Mercury's sellers fall?
Atlanticus now expects to owe $14.0 million, against $40.0 million in December. The payout grows when Mercury's write-offs come in below agreed levels, so a shrinking payout sits oddly beside Howard's "better-than-modeled" claim.
How large are the store-card loans left out of late rates?
Atlanticus leaves out loans whose losses retailers cover from its store-card late-payment rates. Without their size, readers cannot tell how far the reported rates understate late payments across the book.