The Affirm Card supplied 15 of the quarter's 36 points of GMV growth

Strip the card out and the growth rate loses a third of itself. Card GMV of $2.8bn grew 124% ^[13], which puts the year-ago card figure near $1.25bn, and total volume of $14.1bn grew 36% ^[2], which puts the year-ago total near $10.4bn; subtract one from the other at both dates and the non-card business went from roughly $9.1bn to $11.3bn, or about 24% ^[5]. So the Affirm Card, at a fifth of volume ^[7], carried close to fifteen of the thirty-six points ^[6]. That is the product that goes where a credit card goes, which is why the displacement claim from Citizens' David Scharf ^[9] and from CFO Rob O'Hare ^[10] is most literally true of the smallest and fastest piece of the company.

The merchant funnel explains part of that mix. Eighty of the top 250 e-commerce sites is 32% coverage ^[16] against 571,000 merchants signed ^[12], so incremental volume is arriving through consumers carrying a card off-platform rather than through the largest checkouts adding a button. Run the cardholder cohort forward and it looks larger still: 5.2m cardholders at about $4,000 a year is $20.8bn of annualised GMV, some 37% of the run rate ^[10], though the cardholder count doubled during the year, so that figure flatters what they actually contributed.

The guide is where the arithmetic gets specific. A floor of $64bn ^[5] sits 13.5% above the $56.4bn you get by annualising the fourth quarter ^[1], and against a fiscal year that crossed $50bn ^[8] it implies at most 28% ^[2], which makes the eleventh 30% print ^[1], on the company's own floor, the last one. Guided floors are usually set below what a company expects to deliver, so this number caps downside risk more than it forecasts an outcome.

Margin says something about who is paying for the volume. Revenue less transaction costs of $589m on $14.1bn is 4.18% ^[3], and the guided 4.15% ^[5] is three basis points under what was just delivered ^[3]; revenue meanwhile grew 33% against volume's 36% ^[3], so yield per dollar of GMV fell about 2% ^[4] while contribution per dollar rose. That outcome traces to funding and loss costs; pricing was not the driver, and funding costs are the leg most exposed to somebody else's decisions.

On size, William Blair's Andrew Jeffrey puts US carded spend near $9 trillion ^[15] and Affirm under 2% of it ^[14]. The fiscal 2027 floor of $64bn is 0.71% of $9tn ^[12], and a round one percent means roughly $90bn of GMV, which at the guided pace arrives around fiscal 2029 ^[13]. Durable share gain, then, at a rate that leaves the card industry's income statement mostly intact for several more years.

The number that would settle this is the one this quarter's coverage does not carry: losses. Max Levchin's letter offers "steady credit outcomes" ^[8], a phrase with no number attached to it, and RBC's Daniel Perlin named growth and credit quality as exactly what those two guided figures were being read for ^[6]. This is probably wrong, but if the 4.15% holds while delinquencies normalise, the displacement thesis is real and currently underpriced, and if the margin is being held up by pricing a credit cycle takes back, eleven quarters of streak were a funding trade wearing a share-gain costume. The market split the difference inside a session, touching $85.77 and settling at $80.30 ^[16], which gave back 6.8% of its own good news ^[14].