Affirm told investors to expect more on Saturday, November 7. The buy now, pay later (BNPL) provider raised its full-year growth floor for fiscal 2027 to at least 30%, after a first quarter in which gross merchandise volume climbed 34% year over year.

The company also introduced a new AI underwriting model that it says approves more borrowers while keeping credit losses flat.

The numbers behind the raise

Gross merchandise volume, or GMV, measures the total dollar value of purchases running through Affirm's platform. It grew 34% in fiscal Q1 2027 compared with the same quarter a year earlier. Because Affirm's fiscal year ends in June, the reported quarter corresponds to the calendar third quarter.

That performance gave management enough confidence to lift the floor on its full-year outlook to a minimum of 30%. A guidance floor is the low end of management's expected range, so lifting it commits the company to a higher minimum pace rather than simply nudging a midpoint upward. Early second-quarter trends point to GMV growth in the high-30% range, according to the company's results discussion.

Revenue rose 35% for the quarter, coming in at approximately $1.26 billion to $1.3 billion.

The merchant side of the business looked even stronger. Active merchants jumped 48% year over year to 621,000. For a BNPL platform, merchant counts are a distribution metric: every integration places the financing option in front of shoppers at one more checkout.

A transformer for your credit file

The headline technology this quarter is a new transformer-based underwriting model — the same class of neural network architecture that underpins modern large language models. Transformers read an entire data sequence at once, weighing how each element relates to every other, a property well suited to extracting signals from long behavioral histories such as years of repayment records. The model is trained on 14 years of transactional data and is now running in the United States. Affirm says it delivered a 3.4% incremental conversion lift through higher credit approvals, with credit losses holding flat.

The company expects the model to improve approval rates by roughly 2 to 3 percentage points without adding risk. It said the new system outperformed its traditional underwriting methods. That pairing runs counter to the usual trade-off in consumer lending, where wider approvals tend to lift losses — which is what makes the flat-loss figure the crux of the claim.

Room to run, according to management

CEO Max Levchin and CFO Rob O'Hare laid out a long-term growth target exceeding 30%. They pointed to market expansion as the engine: Affirm's share of e-commerce currently sits under 4%, and the company operates in only four markets — figures that frame the target as share gain and geographic expansion on Affirm's side rather than reliance on e-commerce itself growing.

Beyond online checkout, Affirm is pushing into artificial intelligence integrations and software for offline merchants.

What this means for investors and the BNPL race

According to CryptoBriefing's assessment, for shareholders the raised floor matters more than the beat: a company that raises guidance after one quarter is signaling that it sees demand holding up, not just a lucky stretch. The early Q2 read in the high-30% range adds weight to that signal.

What is harder to copy, the report argued, is a credit engine built on 14 years of repayment behavior. If Affirm's model genuinely approves more shoppers at the same loss rate, that becomes a moat competitors cannot buy off the shelf.

It also changes the conversation with merchants. Affirm can pitch a measurable conversion lift, which is a far easier sale than "trust us, consumers like installments."

The next checkpoint comes with fiscal second-quarter results, covering October through December, which will show whether the high-30s pace and the model's loss performance hold through the holiday shopping season.

This article is a standardized rewrite of reporting published by CryptoBriefing on November 7, 2026. The original report was edited by Diego Almada Lopez.