Key Takeaways
- Visa and Mastercard hold a structural advantage where they benefit regardless of which AI platform wins the protocol war, as payments ultimately flow through their networks
- Visa pursues multi-protocol diplomacy with its MCP server and open ecosystem, while Mastercard differentiates through first-mover live transactions and data-driven B2B services
- Both companies are shifting revenue models from transaction fees toward value-added services (VAS), evolving into "payment hyperscalers" for the agent era
Why Card Networks Are the Hidden Winners of Agentic Commerce
As Google's UCP and OpenAI's ACP wage an intense protocol war, the true beneficiaries of agentic commerce may sit elsewhere entirely. Visa's Chief Product and Strategy Officer Jack Forestell has called agentic commerce "the biggest opportunity that I've seen in my 20-plus years in payment technology."
Behind this statement lies a simple structural fact. No matter which platform an AI agent uses to discover products or which protocol processes the order, the final payment passes through Visa or Mastercard's network. Visa connects 14,500 financial institutions with 175 million merchant locations. Mastercard processes 175 billion transactions annually. With infrastructure at this scale, neither company needs to bet on a platform war winner.
Digital Commerce 360's reporting captures this dynamic clearly. Both companies are rivals yet face the same question: how to authenticate AI agents. Their approaches, however, diverge significantly.
Visa Intelligent Commerce — Laying Payment Rails Through Multi-Protocol Diplomacy
Visa has adopted a strategy of backing no single protocol exclusively. The company supports Google's UCP while partnering with OpenAI's ACP, and collaborates with Stripe, Akamai, and AWS to cover the entire technology stack.
Visa Intelligent Commerce (VIC) is the embodiment of this strategy. Over 100 partners participate globally, with more than 30 building in the sandbox and over 20 agent platforms integrating directly. In December 2025, hundreds of agent-driven transactions were completed without a human pressing the checkout button, marking the first large-scale proof of autonomous purchasing.
VIC's design philosophy distills into a single concept: adding an agent-ready layer on top of existing infrastructure. The Trusted Agent Protocol (TAP) uses RFC 9421-based HTTP signatures to verify agent identity, and merchants need only add middleware to their existing web servers. Rather than building a new payment network, Visa runs AI agents on rails it has already laid.
In 2026, Visa expanded its developer toolkit further. The MCP server release enables AI agents to connect directly to VIC APIs, reportedly reducing development cycles from weeks to hours. The Acceptance Agent Toolkit lets non-technical users generate invoices and create payment links through plain-language prompts. AWS collaboration integrates VIC with Amazon Bedrock AgentCore, Asia-Pacific pilots launched in early 2026, and Europe's Agentic Ready program is now live. Visa's vision is a world where every AI platform, every region, and every developer operates on Visa's rails.
Mastercard Agent Pay — Building Credibility Through Live Transactions
Where Visa attacks through diplomacy and ecosystem breadth, Mastercard chose a different battle plan: accumulating live transaction records worldwide to establish a first-mover position.
In Q3 2025, Mastercard CEO Michael Miebach told analysts during the earnings call that the company's network had completed the "first agentic transaction" in the industry. This statement was symbolic. While Visa spoke of partner counts and sandbox scale, Mastercard differentiated with the fact that real transactions were already running.
The rollout accelerated from there. In Australia, a Commonwealth Bank of Australia debit card purchased cinema tickets from Event Cinemas, and a Westpac credit card booked accommodation in Thredbo. In Korea, an AI agent automatically searched, booked, and paid for a car service from Incheon International Airport to a hotel in Gwanghwamun, Seoul. Pilots with HSBC and DBS in Hong Kong, and CIMB and RHB in Malaysia, are also operational.
The common thread across all these transactions is the Mastercard Agentic Token. Unlike standard network tokens, Agentic Tokens uniquely bind an AI agent to an individual user. Consumers authenticate via Payment Passkeys and delegate purchasing authority to the agent within defined boundaries. The token cryptographically constrains the agent's scope of action, directly embodying the philosophy of Verifiable Intent.
The PayPal partnership adds strategic weight. Integrating Agent Pay into PayPal's wallet opens a path for PayPal's 20 million merchants to connect with Mastercard's agentic payment infrastructure.
The Revenue Model Shift Beyond Authentication Standards
The competition over authentication standards between Visa and Mastercard is covered in a separate article. Here, the focus is on the structural shift in revenue models that lies beyond technical standards.
Traditional card network revenue has been anchored in per-transaction processing fees. But both companies' Q1 2026 earnings already point in a different direction. Visa's value-added services (VAS) grew 28%, with non-GAAP operating expenses rising 16%. CEO Ryan McInerney stated the company aims to be "the infrastructure provider and key enabler in agentic commerce so that every agent interaction is trusted and secure." Mastercard's VAS grew 22% on a currency-neutral basis.
What does VAS mean specifically in the agent era? Visa outlines four growth drivers. First, friction reduction improves transaction success rates, as agents execute optimized payments that reduce cart abandonment and increase overall payment volume. Second, transaction density growth, as AI agents fragment purchases into smaller units, enabling pricing "in hours, minutes, or even seconds rather than months or years." Subscription auto-optimization and dynamic pricing responses generate unprecedented transaction frequency.
Third, B2B payment digitization. In a world where agents handle everything from supplier onboarding to invoicing, reconciliation, and payment execution, paper invoices and manual approvals disappear, dramatically increasing on-network transactions. Fourth, data insights. Visa, analyzing roughly 300 billion transactions annually, is entering a "data-as-a-service" space by providing anonymized, aggregated data to AI agents to improve purchase recommendation accuracy.
Mastercard has taken data monetization a step further with concrete products. The Agent Suite, announced in January 2026, combines insights from 4,000 global advisors with Mastercard's transaction data analytics as a B2B AI agent service. In March, Virtual C-Suite launched, offering SMBs executive-level decision support based on 175 billion annual transactions. The initial Virtual CFO module is distributed through financial institutions and accounting platforms.
What this signals is a transformation of card network business models from "pipes" to "platforms." Not merely processing payments, but monetizing the intelligence that payment data generates. This is precisely why PYMNTS calls Visa a "payments hyperscaler."
American Express and the Next Wave of Entrants
How are other card networks responding to the Visa-Mastercard duopoly?
American Express CEO Stephen Squeri positioned agentic commerce as "the biggest change in shopping since eCommerce" in his March 2026 shareholder letter. The company plans to release an Agentic Commerce Experiences (ACE) developer kit enabling partners to integrate Amex payment capabilities into agentic experiences. Amex is also contributing to EMVCo's card-based agentic payment specifications, co-building authentication standards with Cloudflare, and participating in Google's AP2.
Amex's strategy differs fundamentally from Visa and Mastercard. As a combined network, issuer, and acquirer, the company can integrate consumer, merchant, and issuer perspectives simultaneously. It aims to secure its position in the agent era through premium service and security rather than volume.
Google's AP2 has attracted over 60 participants, with JCB and UnionPay International also advancing agentic payment readiness within this framework. However, compared to Visa and Mastercard, their tokenization infrastructure and agent payment flow development lag behind. Delayed adoption risks concentrating agent-driven transaction volume on Visa and Mastercard networks.
| Aspect | Visa | Mastercard |
|---|---|---|
| Core Platform | Visa Intelligent Commerce (VIC) | Agent Pay + Agent Suite |
| Authentication | TAP (HTTP signature-based identity proof) | Agentic Token + Verifiable Intent |
| Alliance Strategy | Multi-protocol diplomacy (UCP + ACP) | Google co-development + OpenAI/Cloudflare |
| Developer Tools | MCP Server + Acceptance Agent Toolkit | Agent Suite + Start Path expansion |
| B2B Expansion | VAS (Value-Added Services) growth | Virtual C-Suite (AI agents for SMBs) |
| First Live Transactions | Dec 2025 (hundreds) | Q3 2025 (industry first) |
| Regional Rollout | 100+ global partners | US → Australia → Korea → Hong Kong → Southeast Asia |
Practical Implications for E-Commerce Businesses
What does the card networks' strategic shift mean for merchants?
The most critical insight is that indirect adoption via payment providers is currently the most efficient path. Stripe supports both Visa and Mastercard Agentic Network Tokens through Shared Payment Tokens (SPT). Major processors including Adyen, Fiserv, and Nuvei are early adoption partners for TAP and Agent Pay. Start by confirming which networks your payment provider supports and to what extent.
Next, consider the indirect benefits from card network VAS within the broader agentic payment landscape. Improved fraud detection accuracy, streamlined dispute processing, and higher transaction success rates all emerge from card networks' AI investments. Even without direct implementation costs, network-side evolution structurally improves merchant profitability.
However, risks warrant attention. As agents fragment purchases and increase transaction density, interchange fee structures may also change. When a single purchase splits into multiple microtransactions, how fee schedules apply remains an open question without industry consensus.
Summary
While Google and OpenAI's protocol wars dominate headlines, a bird's-eye view of the agentic commerce landscape map reveals the unique positioning of card networks. Regardless of which platform wins, payments flow through Visa or Mastercard. Leveraging this structural advantage, both companies are accelerating their transformation from "pipes" to "intelligence platforms." As protocol war outcomes remain uncertain, card networks that have already begun processing live transactions may emerge as the quiet powerholders of the agent era.




