Agentic commerce: Agentic commerce could shift value toward platforms that control consumer intent, trusted transactions, and merchant infrastructure

Overview

This thematic report examines how AI agents could move consumer commerce from product discovery toward comparison, purchase, payment, and eventually autonomous transactions. Goldman Sachs expects a long, uneven adoption cycle rather than immediate disruption, with value increasingly accruing to participants that combine distribution, trust, merchant participation, transaction infrastructure, and control of consumer intent.

Core views

Goldman Sachs frames agentic commerce as the next possible stage of eCommerce rather than an overnight replacement for existing shopping. Consumers already use AI heavily for discovery, research, recommendations, and price comparison: shopping ranks as the third-leading AI application according to Epsilon, about 44% of online buyers use AI tools for product discovery according to Bain & Co., and Epsilon data indicate that six in ten consumers use AI while shopping. Yet end-to-end autonomous purchasing remains early. The report expects adoption over 3-5+ years and potentially over decades, drawing an analogy with eCommerce's gradual development through information discovery, trust-building, lower access barriers, improved delivery, and secure payments. Lower-risk, recurring, and commoditized purchases should migrate first; subjective or high-consequence categories such as apparel and luxury should be slower because an agent's misinterpretation of intent carries greater risk. The addressable opportunity is meaningful but conditional. Goldman Sachs identifies $2.6tn of near-term US consumer spending in high-likelihood categories and estimates that high- and medium-likelihood baskets represent roughly 60% of retail and services spending, or more than $12.3tn of commerce volume. Its sensitivity analysis suggests that roughly 2% agentic penetration of high-likelihood card-present spending could add about 1 percentage point to eCommerce growth; equivalently, about 50bps penetration of high- and medium-likelihood card-present categories could produce that acceleration. The mechanism is reduced discovery and transaction friction, but the report stresses that consumer habits, category-specific trust, merchant willingness to expose catalogs, and reliable transaction infrastructure will govern the pace. The report maps the technology stack across discovery, conversion, and protocols. At the upper funnel, LLMs can search catalogues, compare products, and identify merchants against preferences such as price, shipping speed, and availability. Retailer-owned agents such as Amazon Alexa and Walmart Sparky may lead near-term adoption because they rely on first-party identity, payment acceptance, loyalty, and existing customer relationships, albeit with narrower inventory breadth. Lower-funnel models include browser agents, embedded merchant tools, agent wallets, and personal operating-system agents. Merchant-facing standards such as MCP and Google's UCP are intended to expose catalogues and support discovery, while payment protocols including Visa Trusted Agent Protocol, Mastercard Agent Pay, and AP2 seek to provide authentication, consent, credential binding, and authorization. The report believes that technical protocols establish a path to transact, but clear allocation of responsibility for erroneous purchases, fraud, disputes, returns, and chargebacks remains necessary before broad adoption. This raises a central strategic conflict: horizontal agents may improve discovery for smaller merchants and Shopify's ecosystem, while scaled retailers such as Amazon and Walmart may also benefit because agents optimize for price, availability, delivery speed, assortment, and transaction reliability. Retailers risk losing first-party data, loyalty engagement, cross-selling, retail-media economics, and control of the storefront if independent agents become the primary interface. Amazon's blocking of Meta's Muse from Amazon.com illustrates this tension. Goldman Sachs considers it too early to determine whether open or closed ecosystems will prove more defensible. Agentic commerce may democratize discovery and fragment baskets across merchants, but it may also add new referral, commission, advertising, or placement costs if agent platforms become the new gatekeepers of demand. In advertising, Goldman Sachs expects monetization to migrate with consumer intent rather than vanish. AI is already improving creative production, targeting, measurement, campaign optimization, and return on ad spend. Longer term, if discovery shifts from traditional search or retailer websites to AI interfaces, advertising could move toward sponsored recommendations, catalog-linked formats, and other AI-native placements. The report compares the possible shift to desktop-to-mobile: rather than an app install, the valuable asset may become being the default merchant, service, or brand that an agent repeatedly selects. META and GOOGL are viewed as the principal secular beneficiaries in digital advertising because of their distribution, user relationships, data, AI capabilities, and commerce integrations. META's Muse had more than 2.8 million downloads since launch and became the top-ranked free iOS app in the US in its second week, but account connectivity, permissions, and third-party merchant participation remain important scaling hurdles. GOOGL is viewed as advantaged by its AI stack and integration of Gemini, AI Overviews, and AI Mode. The report sees Shopify as a particularly well-positioned commerce-infrastructure beneficiary. It argues that Shopify's value is not limited to the consumer storefront: its catalogue management, inventory synchronization, checkout, Shop Pay, fulfillment, customer data, order management, identity, and post-purchase workflows remain necessary regardless of whether discovery starts in search, social media, marketplaces, or AI agents. Shopify's work on UCP and Agentic Storefronts could make its integrated merchant platform more valuable as commerce fragments across AI surfaces, while agent-led discovery could improve visibility for independent and emerging brands. Conversely, consumer staples and other low-involvement, easily substituted categories could face greater price comparison, private-label substitution, and commoditization pressure. Goldman Sachs sees more personalized categories, including prestige beauty, as relatively resilient; brands may need to make product information readable and accurately represented by AI systems through a transition from traditional SEO toward generative-engine optimization. Payments and security are core enabling layers. Goldman Sachs believes agentic payments will build on the existing card ecosystem rather than require an entirely new payment form factor. Visa and Mastercard are considered well positioned through tokenization, authentication, fraud prevention, real-time decisioning, global network effects, and potential greater value-added-services attachment. More than 50% of V/MA eCommerce transactions are tokenized, and Visa noted roughly 54% of card-not-present transactions were tokenized, with an estimated approximately $650mn reduction in fraud in FY24. Agentic commerce could also fragment purchases into more, smaller transactions, potentially increasing network yield through fixed transaction fees. Goldman Sachs estimates that a 5% acceleration in V/MA cybersecurity-related value-added services would accelerate VAS growth by 90bps for Visa and 200bps for Mastercard, and total net revenue by 20bps and 80bps, respectively. Modern processors such as Stripe and Adyen could benefit from rising integration complexity, while BNPL providers including AFRM and KLAR could benefit if agents make financing comparisons easier at checkout. The report expects fraud, cybersecurity, identity, and decisioning demand to rise because merchants will have to distinguish authorized agents from malicious automation rather than simply block bots. Device reputation, IP location, navigation patterns, typing cadence, and session velocity may become less useful alone when legitimate agents operate from shared cloud infrastructure at machine speed. Goldman Sachs estimates that cybersecurity spending currently represents roughly 1-2% of eCommerce revenue under its assumptions and expects higher intensity over time as machine-to-machine risks grow. NET could gain through bot management, API security, and application security, although the report says the incremental revenue opportunity is unclear. EFX, TRU, and FICO offer longer-duration optionality around higher-risk events: new account linkages, unfamiliar merchants, financing requests, or material changes in delegated authority. EFX has broad consumer and merchant-verification exposure; TRU is positioned in persistent identity linkage and digital-network risk; and FICO has its clearest potential path through increased automated fraud and financing decisioning in Software rather than incremental Scores volume. In live-event ticketing, Goldman Sachs expects primary platforms to fare better than secondary marketplaces. Agentic discovery could improve personalized recommendations, conversion, and event sell-through, particularly for events with excess capacity; Live Nation has stated that 95% of concerts do not sell out, with amphitheater sell-through of 60-70% and theater sell-through of 65-75%. However, secondary marketplaces with commoditized inventory may face greater price transparency, service-fee pressure, and risk that AI platforms capture a new distribution toll. Primary platforms with differentiated inventory and exclusive supply relationships, such as Ticketmaster, are seen as better positioned. Across sectors, the report's conclusion is that durable winners will be determined less by AI model capability alone than by their ability to capture intent while supplying trusted, scalable, integrated commerce experiences.

Analysis framework

Goldman Sachs starts with the current state of AI shopping adoption and the technical progression from discovery to autonomous purchasing. It then sizes potential US spending exposure, uses historical eCommerce adoption as an analogue, evaluates transaction-risk differences by category, and traces how changes in discovery, authorization, payments, fraud, identity, and distribution could affect advertising, retailers, commerce infrastructure, financial technology, and ticketing.

Methodology notes

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Agentic-commerce ecosystem mapping

  The report follows the flow from consumer discovery to merchant catalogue access, payment authorization, fraud controls, identity verification, financing, and fulfillment to identify which layers could gain or lose value.

• Industry AnalysisVolume-price decomposition

  Agentic-commerce penetration sensitivity

  Goldman Sachs estimates the potential effect on eCommerce growth by applying assumed agentic penetration rates to high- and medium-likelihood card-present spending categories.

• Valuation methodsDCF (Discounted Cash Flow)

  Company valuation approaches

  For selected covered companies, the report cites discounted-cash-flow methods alongside earnings or enterprise-value multiples to support existing price targets.

Asset mapping & comparison

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

• Meta Platforms (META)

  Potential beneficiary through consumer-agent distribution, Muse, engagement, commerce, and advertising monetization.

  Strengths

  Large consumer distribution, established user relationships, advertising infrastructure, and Muse's early traction.

  Weaknesses

  Scaling requires account connectivity, permissions, and third-party ecosystem participation.

  Comparison

  Along with GOOGL, identified as a primary secular beneficiary in digital advertising.

  Risks

  Competition for users and advertising dollars, investment pressure, regulation, and difficulty monetizing optionality.

• Alphabet (GOOGL)

  Potential beneficiary through conversational search, Gemini, AI Mode, and aggregation of commercial intent.

  Strengths

  AI infrastructure, compute, model capabilities, broad distribution, and advertiser and merchant connections.

  Weaknesses

  Discovery migration could require continued evolution of search monetization.

  Comparison

  Alongside META, considered the report's primary digital-advertising beneficiary.

  Risks

  Product and advertising competition, search disruption, investment intensity, and regulatory scrutiny.

• Shopify (SHOP)

  Potential beneficiary as the merchant infrastructure layer for fragmented AI-driven commerce.

  Strengths

  Catalogue, inventory, checkout, Shop Pay, fulfillment, customer-data, and post-purchase infrastructure; UCP and Agentic Storefronts.

  Weaknesses

  Investor debate centers on whether AI agents could disintermediate the consumer storefront.

  Comparison

  The report contrasts Shopify's open integration approach with more restrictive marketplace approaches.

  Risks

  Consumer weakness, investment-related margin dilution, and competition.

• Visa (V) and Mastercard (MA)

  Potential beneficiaries from faster eCommerce growth, tokenization, authentication, fraud prevention, and value-added-services attachment.

  Strengths

  Global network effects, existing card credentials, tokenization, real-time authorization, and agent-payment protocols.

  Weaknesses

  Liability rules and stakeholder roles for agentic transactions remain unsettled.

  Comparison

  Viewed as better positioned than new payment form factors because agentic payments are expected to build on existing card infrastructure.

  Risks

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

• Equifax (EFX), TransUnion (TRU), and Fair Isaac (FICO)

  Potential longer-duration beneficiaries from additional identity, merchant-risk, fraud, credit, and financing-decision activity.

  Strengths

  EFX offers broad identity and merchant verification; TRU offers persistent identity and digital-network intelligence; FICO offers governed fraud and credit decisioning.

  Weaknesses

  The opportunity depends on whether agentic activity creates incremental paid decisions rather than shifts existing workflows.

  Comparison

  FICO is seen as having greater opportunity in decisioning and fraud Software than Scores; EFX has the broadest exposure; TRU is differentiated in identity linkage.

  Risks

  Uncertain revenue conversion from agentic commerce and company-specific macro, credit, competition, and execution risks.

• Cloudflare (NET)

  Potential beneficiary from greater need to verify shopping agents and protect merchant websites and APIs.

  Strengths

  Bot management, API security, application security, and protocol integrations.

  Weaknesses

  The report says the incremental revenue opportunity remains unclear.

  Comparison

  Positioned as a security-enforcement layer alongside payment-network and identity infrastructure.

  Risks

  Uncertain monetization of the agentic-commerce use case.

• Live Nation Entertainment (LYV) and StubHub Holdings (STUB)

  Agentic discovery may improve ticket discovery and sell-through, but effects differ between primary and secondary marketplaces.

  Strengths

  Primary ticketing benefits from differentiated inventory and supply relationships; AI can improve recommendation relevance.

  Weaknesses

  Secondary marketplace inventory is more commoditized and exposed to price transparency.

  Comparison

  Primary ticketing platforms are viewed more favorably than secondary platforms.

  Risks

  New AI distribution fees, service-fee pressure, lower ancillary attachment, and reduced advertising monetization.

Key data

• Near-term addressable US spending$2.6tnHigh-likelihood consumer spending categories identified for agentic commerce.
• High- and medium-likelihood commerce volumeMore than $12.3tnRoughly 60% of current retail and services spending under the report's categorization.
• Potential eCommerce growth accelerationApproximately 1 percentage pointEstimated from roughly 2% agentic penetration of high-likelihood card-present spending.
• AI product-discovery usage~44% of online buyersBain & Co. estimate cited by the report.
• Retail media market~$170bn globally in 2025Statista estimate cited by the report, versus $77bn in 2020 and a 17% five-year CAGR.
• Meta Muse downloadsMore than 2.8 millionSince release; the app became the top-ranked free iOS app in the US in its second week, according to Sensor Tower.
• Cybersecurity intensity~1-2% of eCommerce revenueImplied by the report's assumptions on retailer cybersecurity budgets and online operations.

Impact & implications

The report sees agentic commerce as a potential source of incremental eCommerce growth and a redistribution of value toward platforms with distribution, trusted consumer relationships, merchant integrations, secure payment rails, and risk-management capabilities. It also highlights risks to merchant customer ownership, retail-media economics, low-involvement branded categories, and secondary ticketing platforms with commoditized inventory.

Risks

• Consumer adoption may be gradual because shopping habits, trust, and category-specific tolerance for purchase errors change slowly.
• Liability for fraud, chargebacks, returns, and erroneous purchases remains unresolved and may limit merchant and consumer participation.
• Merchants may restrict third-party agent access to preserve customer ownership, first-party data, loyalty, and retail-media economics.
• AI platforms may become new distribution gatekeepers, charging referral, placement, commission, or advertising fees.
• Authorized agents may weaken the usefulness of traditional fraud signals and increase exposure to malicious automated activity.
• Price transparency and agent-led comparison may pressure margins, branded differentiation, service fees, and ancillary revenues in commoditized categories.

What to watch

• Whether agents move beyond discovery into reliable end-to-end purchasing and autonomous transactions.
• Development of consumer-intent verification, authentication, fraud-loss, return, dispute, and liability frameworks.
• Merchant willingness to expose catalogues, inventory, checkout, and payment infrastructure to third-party agents.
• The competitive balance between retailer-owned agents, horizontal AI platforms, and open versus closed commerce ecosystems.
• Adoption of UCP, MCP, Visa Trusted Agent Protocol, Mastercard Agent Pay, AP2, and related tokenization standards.
• Whether AI-native advertising and sponsored recommendations replace or supplement search and retail-media monetization.
• Whether agent-led activity creates incremental paid verification and decisioning volume for EFX, TRU, and FICO.