Mastercard’s AI Payment Pilot Puts Pressure on Visa
Santander processed what the companies describe as Europe’s first AI-agent transaction. Visa is betting on network reach and standards instead
Mastercard and Banco Santander say they have completed what they describe as Europe’s first live, end-to-end payment executed by an AI agent. The transaction was processed using Mastercard’s Agent Pay framework.
Santander says the AI-initiated payment ran inside the bank’s regulated payment framework and within predefined customer limits as part of a controlled pilot rather than a commercial launch, according to a press release.
The European payments system operates under the European Union’s revised Payment Services Directive, known as PSD2. Under PSD2, payment service providers must apply strong customer authentication when a payer initiates an electronic transaction, requiring at least two independent authentication factors, according to guidance published by the European Banking Authority.
Any AI-initiated transaction would need to comply with those authentication and liability requirements.
Matías Sánchez, global head of cards and digital solutions at Santander, said in a statement: “Our role is not only to adopt innovation, but to shape it responsibly, embedding security, governance and customer protection by design. As AI agents become part of everyday commerce, building trusted, scalable frameworks will be essential to unlocking their full potential.”
Mastercard and Visa Take Different Paths on AI Payments
Mastercard introduced Agent Pay as infrastructure that allows AI systems to initiate and complete payments on behalf of users, according to company materials.
The Santander pilot places that capability inside a regulated banking environment, where authentication, transaction limits and liability rules are already defined under PSD2. The test demonstrates that an AI agent can execute a compliant payment within existing European regulatory guardrails.
Visa is approaching the problem from a different layer. It has launched Intelligent Commerce and the Trusted Agent Protocol, aimed at creating a standardized method to verify AI agents across merchants and issuers.
The company has also expanded its fraud and anomaly-detection capabilities through its acquisition of Featurespace, positioning those tools as part of the control framework for agent-initiated transactions.
Mastercard’s pilot tests execution inside a bank’s regulated flow. Visa’s effort focuses on identity verification, interoperability and fraud controls at network scale.
Still, Visa has said in press materials that it has completed agent-initiated transactions with partners. “We are seeing impressive progress in how AI will transform commerce, with many real-world transactions completed by Visa’s deep network of partners,” Rubail Birwadker, Visa’s head of growth products and partnerships, said in a company statement.
Read more: Visa Just Completed Hundreds of AI-Initiated Purchases
Meanwhile, publicly available reporting does not describe a comparable live AI-agent payment processed through another European bank.
The competitive stakes are reflected in the companies’ relative size. In fiscal 2025, Visa processed approximately 257.5 billion transactions representing roughly $14.2 trillion in payment volume, according to Visa’s 2025 annual report. Mastercard processed approximately 175.5 billion switched transactions and recorded about $10.6 trillion in gross dollar volume across its payment network, according to Mastercard’s filed results.
Industry benchmarks show Visa as the largest global card network outside China, accounting for about 50% of total card payments worldwide. Nilson Report-based purchase volume data show Visa at roughly 32% of global credit card purchase volume and Mastercard at about 21%, indicating Mastercard’s share is materially smaller on that measure.
AI adoption in commerce is already broad. About 89% of retailers are using or testing AI daily or through pilot programs, and roughly half of e-commerce businesses have adopted AI technologies in their operations.
As AI agents begin interacting directly with payment systems, the question becomes whether execution inside regulated banks or network-wide verification standards will shape how those transactions scale.