Mastercard introduces “Agent Pay for Machines” to let AI systems pay each other

Mastercard has introduced Agent Pay for Machines, a new offering designed to enable AI-driven and connected systems to make payments to each other automatically. The company announced the launch on June 10, positioning the service as a way to extend payment capability beyond people and traditional applications to “agents” that can act on behalf of users or organizations in digital environments.

The concept is straightforward: when two systems need to exchange value, Agent Pay for Machines is intended to allow the payment to be initiated and processed through Mastercard’s payment rails without manual steps. In the same way that online payments have become increasingly automated for software platforms, Mastercard’s move targets a future where software and connected devices are continuously negotiating, ordering, and settling transactions with minimal human involvement.

Mastercard’s announcement frames the initiative as an “AI-powered” approach to machine-to-machine payments. That language suggests the company is targeting scenarios in which AI agents decide when to pay, how much to pay, and potentially what payment method to use, based on the context of a transaction. Mastercard did not, in the available announcement text, spell out the scope of supported AI use cases or how agents would be authenticated and authorized in practice.

The company also appears to be focusing on cross-border scale by referencing payments “across its global” network in the announcement description. Mastercard did not provide further detail in the available information about coverage by country, supported payment types, settlement timing, or whether the service supports cards, bank transfers, or other rails for machine-to-machine payments.

It is part of a broader push in the industry to treat payments as software infrastructure. As organizations experiment with autonomous workflows, the ability to pay reliably, reconcile quickly, and comply with financial rules becomes more important than the ability to accept a one-time payment from a human. For Mastercard, adding another layer that can sit between AI systems and payment processing could help keep it involved as transactions become more automated and more programmable.

At the same time, the move raises unanswered questions that typically matter for automated payments. Mastercard did not disclose in the available post who the first customers are, what partners or platforms integrate with the service, or what controls exist to prevent erroneous or fraudulent payments initiated by malfunctioning AI. In addition, pricing, contractual terms, and service-level details are not included in the information available here.

For the market, the announcement indicates that major payment networks are taking the “agent economy” seriously and are moving from pilots and concepts toward named products. If Agent Pay for Machines works as advertised, it could reduce friction for AI-mediated commerce, streamline vendor payments triggered by systems automatically, and create new opportunities for payment-driven platforms to embed settlements into workflows.

What to watch next is whether Mastercard publishes more technical and commercial specifics, such as integration requirements, supported geographies and payment rails, and examples of real deployments. Traders and customers will also want clarity on how Mastercard handles security, permissions, and dispute resolution when payments are triggered by software rather than people. Until those details are available, the launch should be viewed as a product announcement rather than a fully specified implementation roadmap.