Agentic commerce in payments and fintech: Goldman Sachs sees agentic commerce as a long-term payments catalyst, led by card-network tokenization, security services, and BNPL

Overview

This payments-and-fintech report tracks investor debate following Goldman Sachs' agentic-commerce deep dive. Its central conclusion is that AI-enabled shopping could expand e-commerce activity, with card networks, security providers, modern processors, and BNPL firms positioned to benefit through different mechanisms.

Core views

Goldman Sachs views agentic commerce as a potential accelerator of e-commerce payment volumes and a catalyst across the payments ecosystem. Investor feedback on Visa and Mastercard was broadly positive, especially compared with the more bearish discussion earlier in 2026. The principal uncertainty is the timing of a transition from current implementations toward network-led agentic tokenization infrastructure. For card networks, the report argues that global network effects make card payments the simplest way to enable agentic commerce. Visa and Mastercard could participate in volume growth and see transaction-yield support from basket fragmentation and lower average order values, alongside a higher attachment of value-added services. Goldman Sachs expects most agentic transactions eventually to use Visa Token Service and Mastercard Digital Enablement Service. It also expects the enlarged fraud and identity-verification challenge to create demand for network security products. The report estimates that a 5% acceleration in Visa and Mastercard cybersecurity-related value-added services would accelerate VAS growth by 90 basis points and 200 basis points, respectively, and total net-revenue growth by 20 basis points and 80 basis points, respectively. The report describes Meta's September 8, 2026 launch of Muse, a personal agent that can browse, perform tasks, and make purchases. Its payments currently rely heavily on Stripe infrastructure and Stripe Link's Wallet for Agents, using either Stripe token infrastructure or a single-use virtual card. Goldman Sachs views the virtual-card-heavy setup as a speed-to-market solution rather than a durable technical end state. The two-step consumer-to-platform and platform-to-merchant flow can complicate fraud and chargebacks because Stripe may sit between merchant and consumer; it may also reduce issuer and merchant data, impairing approval rates, merchant loyalty, and issuer rewards. The report expects payment flows to become more streamlined as network tokenization reduces steps. It sees a likely eventual outcome in which Apple Pay and Google Pay are available in Muse, with consumer wallet behavior broadly resembling current mobile-wallet usage and context-specific wallets retaining roles on platforms. The report separates agentic commerce into upper-funnel discovery, lower-funnel conversion, and enabling protocols. AI tools are already affecting discovery and product research, allowing consumers to search inventory, compare products, and apply constraints such as shipping times and price limits; this may reduce the role of conventional search engines. End-to-end purchase and payment functions remain earlier-stage, appearing through browser agents, module-based agents, embedded wallets, and personal operating-system agents. Wider adoption still requires clear rules on stakeholder responsibilities and liabilities. Goldman Sachs outlines ecosystem infrastructure including MCP for secure access to merchant data and inventories, Google's Universal Commerce Protocol and Agent Payments Protocol, Stripe's Agentic Commerce Protocol, and Visa and Mastercard payment protocols. Stripe Link can use Shared Payment Tokens for Link-enabled merchants and a single-use virtual card for other merchants after consumer approval. Visa Trusted Agent Protocol and Mastercard Agent Pay are intended to verify agents, track user intent and purchase mandates, and depend on network tokenization. More than 50% of Visa and Mastercard e-commerce transactions are already tokenized, according to the report. Tokenization keeps underlying payment credentials from being exposed and, in traditional payments, has reduced online fraud and improved authorization rates. For processors, Goldman Sachs believes agentic commerce adds complexity to payment processing and creates a new basis for judging authorization rates and integration quality. It expects this to reinforce existing competitive dynamics favoring modern e-commerce processors such as Stripe and Adyen over processors built through numerous acquisitions, including Global Payments and Fiserv. It does not view agentic commerce as intrinsically negative for PayPal because wallet choice should still guide volumes, but it does not expect PayPal's competitive position to improve without structural changes to its merchant and consumer proposition. PayPal's two-sided ecosystem has near-term strategic value because agents and credentials must be tied to the end consumer, and its enablement in Meta Muse is consistent with that view. For consumer fintech, the report expects consumers to retain their usual preferred credit-card or debit-card payment choices, but believes agents may make it easier to select lower-cost and simpler financing at checkout. That could support BNPL providers, especially Affirm and Klarna, as agentic surfaces develop through wallet integrations and eventually direct integrations. Goldman Sachs acknowledges investor concerns over the absence of direct BNPL integrations in Muse, but views the additions of PayPal and Shop Pay as evidence that payment infrastructure is still being built. It believes direct BNPL integration could follow because credit availability and conversion optimization are incentives. The lack of effectively free upper-funnel advertising on merchant sites could modestly pressure BNPL acquisition, but Goldman Sachs expects BNPL to benefit when consumers ask agents to locate the best credit option.

Analysis framework

Goldman Sachs starts with investor feedback and then evaluates payment-flow design, existing agentic-commerce implementations, protocol infrastructure, security needs, and the competitive effects on networks, processors, wallets, and BNPL providers. It links operational mechanics—such as tokenization, authorization, fraud liability, and wallet integration—to potential volume, service-revenue, and competitive implications.

Methodology notes

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Agentic-commerce payment ecosystem analysis

  The report traces how AI shopping agents, merchant integrations, wallets, processors, card networks, issuers, and BNPL providers interact and how changes in payment architecture can transmit to each participant.

• Industry AnalysisVolume-price decomposition

  Transaction-volume and yield analysis

  Goldman Sachs links potential e-commerce volume growth, lower average order values, and value-added-services attachment to possible payment-network revenue effects.

• Valuation methodsP/E and PEG Valuation

  Forward earnings multiples for Visa and Mastercard

  Visa's $438 target uses a 27x Q5-Q8 P/E multiple, while Mastercard's $701 target uses an approximately 28x multiple on Q5-Q8 EPS estimates.

• Valuation methodsEV/EBITDA valuation

  Enterprise-value-based valuation for Affirm and Klarna

  Affirm's $115 target uses an approximately 21x EBIT less stock-based-compensation multiple, while Klarna's $22 target uses an approximately 14x EV multiple on adjusted operating income less stock-based compensation.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Visa Inc. (V)

  Card-network beneficiary of agentic-commerce volume, tokenization, and security-service demand.

  Strengths

  Visa Token Service, a broad risk-and-security portfolio, and potential yield benefits from lower average order values.

  Weaknesses

  The timing of agentic-tokenization adoption remains uncertain.

  Comparison

  Positioned alongside Mastercard as a leading network-tokenization provider.

  Risks

  Weaker macro conditions, slower cross-border recovery, competition, and unfavorable regulation.

• Mastercard Inc. (MA)

  Card-network beneficiary of agentic payments, value-added services, and fraud-prevention demand.

  Strengths

  MDES tokenization, product investments and acquisitions in security, and a claimed 40% more fraud detection from AI-powered Decision Intelligence.

  Weaknesses

  Near-term competition from processors, banks, and technology companies developing their own frameworks.

  Comparison

  Positioned alongside Visa; Goldman Sachs highlights Mastercard particularly for fraud and identity-verification capabilities.

  Risks

  Weaker macro conditions, slower cross-border recovery, competition, and regulation.

• Affirm Holdings (AFRM)

  BNPL provider that may benefit as agents help consumers select financing at checkout.

  Strengths

  Potential direct and wallet-based integration paths, including existing Apple Pay and Google Pay links.

  Weaknesses

  Near-term absence of direct integration in Muse and potential pressure on upper-funnel customer acquisition.

  Comparison

  Highlighted with Klarna as a modern credit issuer positioned to gain share.

  Risks

  Worsening macro conditions, increasing competition, and unfavorable regulation.

• Klarna Group Plc. (KLAR)

  BNPL provider that may benefit from consumer demand for cheaper and simpler point-of-sale financing.

  Strengths

  Potential wallet and direct-integration routes as agentic payments mature.

  Weaknesses

  Customer-acquisition outcomes depend on checkout design and the balance between wallet and direct integrations.

  Comparison

  Highlighted with Affirm as a potential BNPL beneficiary.

  Risks

  Underwriting risk, worsening competition, slower-than-expected BNPL adoption, merchant pressure on MDRs, and less favorable regulation.

• PayPal Holdings (PYPL)

  Wallet participant with near-term strategic value in agentic commerce through its consumer and merchant ecosystem.

  Strengths

  Two-sided ecosystem and recent enablement within Meta Muse.

  Weaknesses

  Goldman Sachs does not expect competitive dynamics to improve absent structural changes in positioning and value proposition.

  Comparison

  Competes for wallet relevance against mobile and context-specific wallets.

Key data

• Tokenized Visa/Mastercard e-commerce transactions>50%The report states that more than half of V/MA e-commerce transactions are tokenized.
• Cybersecurity VAS acceleration scenario5%A 5% acceleration in cybersecurity-related VAS is estimated to add 90bps/200bps to Visa/Mastercard VAS growth.
• Visa total net-revenue growth effect20bpsEstimated acceleration from the 5% cybersecurity-VAS scenario.
• Mastercard total net-revenue growth effect80bpsEstimated acceleration from the 5% cybersecurity-VAS scenario.
• Visa 12-month price target$438Based on a 27x Q5-Q8 P/E multiple.
• Mastercard 12-month price target$701Based on an approximately 28x Q5-Q8 EPS multiple.
• Affirm 12-month price target$115Based on an approximately 21x EBIT less SBC multiple on Q5-Q8 estimates.
• Klarna 12-month price target$22Based on an approximately 14x EV multiple on Q5-Q8 adjusted operating income less SBC.

Impact & implications

The report sees agentic commerce as strengthening the strategic role of tokenization, authentication, fraud prevention, authorization optimization, and wallet connectivity. It expects the economic benefits to vary by position in the payment stack: networks may gain from volumes and security services, modern processors from integration quality, and BNPL firms from financing choice at checkout.

Risks

• Rules and liabilities across agentic-commerce stakeholders remain unresolved and may delay broader adoption.
• Current virtual-card-heavy flows can create fraud, chargeback, approval-rate, loyalty, and rewards challenges.
• Network tokenization may take time to proliferate, while processors, banks, and technology companies develop competing frameworks.
• For BNPL, direct-integration timing and the loss of upper-funnel merchant-site advertising could affect customer acquisition.

What to watch

• The pace at which current virtual-card implementations shift toward network agentic tokenization.
• Whether Apple Pay, Google Pay, and BNPL providers receive broader wallet or direct integrations in agentic platforms.
• How fraud liability is assigned in agentic-commerce transactions.
• Authorization rates, consumer wallet choice, and merchant/issuer data retention as agentic payment flows scale.
• Adoption of protocols including Stripe ACP, Google UCP/AP2, Visa VTAP, and Mastercard Agent Pay.