JPMorgan Report: AI Shopping Agents Use Card Networks; Payment Impact Limited

Written by: Rita

Six months ago, the market expected agent commerce to require complex merchant integrations and alignment of agent agreements. Now, Meta’s Muse enables payments for non-integrated merchants via a single virtual card and for integrated merchants via tokenized cards, making the payment flow resemble traditional e-commerce. JPMorgan believes that the order in the payments processing world is increasingly returning to normal, with limited near-term impact on covered companies.

In a payment processing industry report released on October 23, 2026, J.P. Morgan noted that Muse partners with Stripe Link, directing non-integrated merchants through virtual card dual transactions and integrated merchants through tokenized card transactions. This mechanism resembles the early non-integrated payment flows in food delivery and buy-now-pay-later, resulting in dual transactions in the short term while remaining primarily card-based infrastructure in the long term. Payment networks benefit; online checkout faces competition, but trust and lending moats remain intact. The competitive landscape for traditional acquirers is more favorable than previously anticipated, with neutral impacts from new banks and remittance companies.

The adoption rate of agency commerce, the dual-transaction mechanism of payment networks, the competitive landscape of online checkout, and the evolving environment for traditional acquirers are four dimensions for understanding recent trends in payment stocks. Stock preferences have diverged, with overweight ratings concentrated on Affirm, Chime, Marqeta, Remitly, and Toast; neutral ratings for PayPal and Klarna; and underweight for Western Union.

Muse’s payment flow is key to understanding the evolution of agency commerce. Non-integrated merchants receive payment via single-use virtual cards: Stripe Link issues cards to consumers, and the virtual cards then pay the merchant, resulting in two card network transactions. Integrated merchants use the consumer’s underlying funds directly, primarily through tokenized card transactions.

Six months ago, the market expected agent commerce to require complex merchant integrations and alignment of agent agreements. Recent advancements from Muse and Instinct show that browser checkout is emerging as the standard for early agent commerce. Browser checkout does not require merchant integration; agents collect product information via web pages and check out on behalf of users. While less reliable and efficient than headless checkout, it offers faster deployment.

JPMorgan Chase noted that responsibility allocation remains a key area of focus. For card-present transactions, the issuing bank typically bears liability for fraud and chargebacks, while for card-not-present transactions, the merchant typically bears liability. Liability allocation for proxy transactions has not yet been determined; adoption of proxy commerce could accelerate if any entity is willing to assume liability as the recorded merchant.

The dual-transaction mechanism for non-integrated merchants provides short-term benefits to Visa and Mastercard. Each agency transaction typically generates two card network transactions and fees: one for the virtual card top-up and another for the virtual card payment to the merchant. Transactions with integrated merchants, by contrast, are primarily tokenized card transactions, and token usage on the network is expected to increase over the long term.

Consumers strongly prefer funding proxy transactions with cards or wallets. According to J.P. Morgan’s 2026 Online Checkout Survey, only 3% of consumers prefer using bank accounts to fund e-commerce transactions. Proxy transactions involve unknown counterparties, making the fraud protection and dispute rights offered by cards more valuable.

JPMorgan believes that the central role of payment networks in agency commerce has not changed. Short-term dual transactions are transitional gains, while long-term tokenized transactions are a more stable source of revenue. Agency commerce does not bypass traditional payment rails; card-based infrastructure remains core.

Agency commerce poses a challenge to online checkout providers, with the most direct risk being friction monetization. Consumers choose a particular checkout button partly for convenience, and agencies can absorb this friction, potentially reducing button usage. J.P. Morgan notes that trust and lending moats still exist.

Online checkout surveys show that about 30% of consumers are at least somewhat comfortable with proxy transactions, with approximately 50% preferring to fund these transactions via digital wallets—PayPal leads with a 32% preference. J.P. Morgan believes that the brand and trust advantages of digital wallets remain effective in proxy commerce, and the wallet competition may intensify.

The moat for lending is harder to replicate. As intermediaries between consumers and merchants, agencies face increased underwriting complexity, giving technologically capable providers a first-mover advantage. Affirm is well-positioned in terms of underwriting complexity, fee structure, and consumer trust; Klarna and PayPal have brand and trust advantages but must demonstrate their ability to maintain market share in agency commerce.

Agency commerce resembles traditional e-commerce, and the competitive landscape for traditional acquirers is better than expected six months ago. J.P. Morgan previously believed that the technical complexity of agency commerce would intensify competition in e-commerce acquiring, favoring new payment service providers with stronger technical capabilities. However, the browser checkout model has reduced integration complexity, making scale and cost advantages important once again.

E-commerce payment processing has consolidated among five providers: Adyen, Braintree (PayPal), JPMorgan Chase, Stripe, and Worldpay (Global Payments). Last year, Stripe, JPMorgan Chase, and Adyen collectively added approximately $1 trillion in transaction volume, about half of Visa and Mastercard’s combined transaction growth of $2 trillion, representing a total transaction volume increase of around 20%.

Global Payments generates approximately 75% of its revenue from card-present transactions and is less affected by competition from agent-based acquiring. Fiserv has only a low single-digit percentage of its revenue exposed to e-commerce acquiring, resulting in limited impact. PayPal’s Braintree has a stronger position in agent-based acquiring, but Braintree accounts for only a low single-digit percentage of PayPal’s transaction processing revenue. Toast focuses on card-present verticals, resulting in limited exposure to agent-based acquiring. Square’s card-not-present acquiring represents a low to mid-single-digit percentage of Block’s gross profit, with manageable impact.

The adoption rate of agency commerce has been a key factor with limited recent impact. Muse has approximately 12 million global downloads, and Instinct has an annualized trading volume of about $1 billion with hundreds of thousands of users. Twenty-five percent of Muse’s users utilize it for shopping, with an average transaction amount of $200.

Approximately 40% of Instinct users add a credit card within the first three weeks, but early adopters exhibit higher trust in AI than the general public. A J.P. Morgan online checkout survey found that only about 30% of consumers feel at least somewhat comfortable with agent-based commerce, which better reflects near-term adoption potential. E-commerce accounts for 17.1% of retail in the U.S., having risen by approximately 9 percentage points over the past decade—changing consumer spending habits takes time.

JPMorgan conducted a sensitivity analysis on user count and average spending per user. Under the scenario of 60 million users with an annual average spending of $6,000, U.S. agency commerce purchase volume would reach approximately $360 billion, accounting for 3.4% of Visa and Mastercard’s U.S. purchase volume in 2026. Under the scenario of 100 million users with an annual average spending of $10,000, the share would rise to 9.3%. Due to the currently small user base and slow adoption rate, the impact on covered companies is limited.

JPMorgan Chase prefers Affirm for online checkout due to its technological complexity and underwriting capabilities, making it suitable for early integration into agency commerce. PayPal’s brand and trust advantages are beneficial in wallet competition, while Klarna also has brand strength but must demonstrate its ability to maintain market share.

On the neobank front, Chime offers low-cost banking services and could benefit as AI assistants help consumers find low-fee banking options. JPMorgan believes Chime’s stock has declined 6% since the release of Muse, underperforming the S&P 500 by approximately 7 percentage points, reflecting an overreaction. Dave faces higher risks in liquidity product pricing, and Cash App’s instant deposit fees may come under pressure, but swap and lending revenues remain unaffected.

In terms of remittance pricing, Remitly and Western Union are positioned at the mid-to-high end compared to peers, and AI assistants may intensify price competition. J.P. Morgan believes that trust and payment network coverage constitute moats; Remitly’s ability to capture market share despite higher prices indicates that trust and reliability outweigh price sensitivity. Remitly’s integration with ChatGPT enhances the discovery process and could drive incremental transaction volume.

JPMorgan coverage ratings: Affirm Buy, Block Buy, Chime Buy, Dave Buy, FIS Hold, Fiserv Hold, Global Payments Hold, Klarna Hold, Marqeta Buy, PayPal Hold, Remitly Buy, Toast Buy, Western Union Sell.

The adoption rate of agency commerce, allocation of transaction liability, and wallet competition will determine the next divergence point for payment tokens.

Disclaimer

This article is a compilation and interpretation by Chaoxiang Research of a third-party brokerage research report (J.P. Morgan, October 23, 2026), combined with publicly available market information. The ratings, price targets, earnings forecasts, and related judgments cited herein reflect the views of the brokerage’s analysts and represent the position of their respective institution only; they do not reflect the views of Chaoxiang Research nor constitute any investment advice.

The market carries risks; make decisions independently. This article should not be used as a basis for buying or selling any securities.