New Research Finds 61% of U.S. Consumers Welcome AI-Assisted Pay Later Recommendations, But on Their Terms

PYMNTS Intelligence (with Splitit) finds consumers are receptive to AI at checkout: 61% of 2,034 U.S. respondents would allow an AI shopping assistant to recommend Pay Later, rising to 80% for Gen Z, but only with control and transparency. Credit-score protection is a baseline requirement (59%), and card-linked installments are favored over standalone BNPL (36% vs 12% used in the past three months), suggesting AI-assisted Pay Later growth will hinge on lowering friction without new credit applications.

Analysis

The durable signal here is not “AI checkout” as such, but that AI becomes a recommendation layer while the payment instrument remains the moat. That structurally favors card networks and issuers that can expose installments inside an existing account relationship, while standalone BNPL names face a tougher distribution fight: AI assistants are likely to compress merchant-side impulse conversions that previously depended on branded BNPL placement. Over 6-18 months, that means lower CAC efficiency and weaker pricing power for pure-play BNPL versus embedded card-linked offerings.

For Alphabet, the opportunity is more about control of intent than payment economics. If AI shopping agents become the default front end, GOOGL can deepen commerce engagement and ad-to-transaction conversion, but the monetization likely leaks to Visa/Mastercard/issuing banks unless Alphabet owns routing or checkout orchestration. The second-order read-through is favorable for payment orchestration and merchant acquisition layers only if they can become the default rails inside AI-driven commerce; otherwise this is mostly a UX story, not a margin story.

The contrarian point is that survey enthusiasm may overstate actual adoption: consumers want AI assistance, but they want approval rights and credit-score neutrality, which keeps the workflow closer to guided search than autonomous purchasing. That makes the near-term catalyst path slower than the headline suggests: 1-3 months of product announcements and pilot integrations, but only modest revenue impact until issuer/merchant systems are actually embedded. The thesis is falsified if standalone BNPL shows accelerating merchant attachment or if card-linked installments fail to lift approval/conversion metrics in issuer disclosures.

Key Decisions for Investors

• Relative-value long V/MA vs short AFRM/KLAR/ZIP over 3-6 months: the survey favors existing-card installments over new-credit BNPL, but keep sizing modest because this is a distribution thesis, not a near-term earnings catalyst.
• Stay constructive on GOOGL on dips, but treat this as an optionality holder rather than a direct payment monetization trade; the cleaner catalyst is any product update showing AI shopping agents can complete transactions inside Google surfaces.
• Watchlist alert on card-issuing proxies (COF, AXP, DFS if liquid) for evidence of installment take-up in next earnings: if managed spend/utilization or interchange trends inflect, that confirms the embedded-credit thesis.
• Avoid chasing long-only Splitit-adjacent or BNPL-adjacent names on the headline alone; wait for merchant attach rates, approval rates, and repeat usage data before underwriting a durable shift.