Affirm Holdings, Inc. AFRM

Key data

The business

Affirm is point-of-sale lending dressed as a checkout button. A shopper buying a $1,400 Peloton or a $60 pair of sneakers picks Affirm at checkout and splits the cost, either four interest-free installments the merchant subsidizes, or longer monthly terms that carry interest the shopper pays. Affirm charges no late fees, ever. It makes money three ways: a fee the merchant pays to offer the button, interest from consumers on longer loans, and gains from selling those loans to investors. Unlike a card network, Affirm holds the credit risk itself, so it is as much a lender as a payments company, and it funds the loans through warehouse lines and securitizations rather than deposits. The moat is a two-sided network reinforced by underwriting data: the more merchants integrate the button and the more repayment history it gathers, the sharper its pricing and the harder it is to dislodge. Its newest wedge is the Affirm Card, a physical card meant to turn an occasional checkout tool into something a consumer reaches for weekly.

The numbers

The trajectory is a company crossing from cash-burning growth into real profitability, with one quarter that needs an asterisk.

Revenue climbs off a seasonal cadence, with the December quarter the peak. Then look at the last row. Net income of $1.62 billion on $1.17 billion of revenue is not an operating result, it is a one-time release of a deferred tax asset, the recognition that years of losses can now shelter future profit. Strip it out and Affirm earned about $500 million pretax for the year. The $4.62 print, which cleared a $0.35 estimate more than ten-fold, is that tax gain, not the business.

Across five years revenue more than tripled, compounding about 33% a year, while the company crossed from an $866 million operating loss in 2022 to a $417 million operating profit in 2026. That swing, not the tax headline, is the story: operating margin went from deeply negative to positive 10% in a single year, the classic inflection of a platform whose costs were mostly fixed and whose revenue finally caught them. On the volume, about $45 billion ran through Affirm, up 38%, and the merchant fee on that, about 3.2 cents on the dollar, held steady rather than eroding, which for a payments business is the whole game. Here is what the market is weighing wrong: at 13 times reported earnings Affirm looks like a cheap fintech that turned the corner, but on the roughly $375 million it would have earned at a normal tax rate the stock sits near 65 times, and the entire case rests on revenue compounding above 30% for years so the forward estimates, 25 times two-year-out profit, converge to reality. The print that settles it is next year's operating margin: keep expanding it and today's price is cheap, stall and it is not.

Management

Max Levchin, a PayPal co-founder, still runs Affirm and remains its public face, which matters in a business whose whole pitch is trust. The capital record is mixed. Affirm bought back $250 million of stock in fiscal 2025 and none in 2026, sensible given the price ran, and it has never paid a dividend. Insider activity is one-directional: twelve sales worth $23 million over the past year and not one open-market purchase. CFO Michael Linford sold $8.0 million in June and another $7.1 million on August 28, the day after the 10-K landed, and the chief legal officer sold $3.1 million days later. Plan status is not disclosed in the filings I read, so I cannot separate scheduled selling from discretionary, though a CFO selling the morning after results is worth a raised eyebrow. Pay leans heavily on stock, which aligns the team with the shares but dilutes, running about 4% of revenue.

How it fails or surprises you

Credit cracks under the held book. Affirm keeps the loans, so a consumer slowdown hits it directly. The tell I have watched is 30-plus-day delinquencies climbing while funding costs rise at the same time, which drops earnings and multiple together. This quarter it held: customer service and collection costs rose 36%, in step with 38% volume, not ahead of it, so credit is keeping pace with growth rather than breaking.

Funding gets expensive. Affirm has no deposits, it borrows to lend. Long-term debt climbed from $7.6 billion to $9.8 billion in a year, up 29%, to fund the growing book. If securitization spreads widen in a risk-off market, the gain-on-sale and net interest that drive revenue compress fast, and the volume that reads as strength becomes a liability.

The Affirm Card becomes a daily habit (right tail). The market pays for BNPL at checkout, a spiky, occasional use. It does not pay for Affirm becoming a card a consumer swipes weekly. If card adoption and revenue per active consumer inflect, revenue reaccelerates past the 30% baked into estimates, and the first place it shows is card volume outrunning overall volume.

Closing thoughts

Operating margin over the next year settles whether the inflection is real. Affirm is a genuine platform inflection wearing a distorted headline: the operating turn is real and accelerating, but the reported earnings that make it look cheap are a tax artifact that will not repeat. The fatter tail is arguably the upside, because a margin that moves from -3% to +10% in one year rarely stops at 10%, and the Card gives revenue a second engine the estimates barely credit. The left tail is equally clear and cheap to watch: a credit and funding squeeze arriving together, the one thing that takes earnings and multiple down at once. What is at risk if that breaks is real, this is a levered lender, but the evidence this quarter says it is not breaking yet.

The bet is still that Affirm signs up more shoppers, gets more big stores to offer it, and keeps getting paid back. What breaks it is the held loan book going bad while borrowing gets dear at the same moment, and the two numbers that tell you first are 30-plus-day delinquencies on that book and the spread Affirm pays on new securitizations. Watch those two together, the tax-flattered $5.53 will not cushion you if they turn.