Mastercard unveils “Agent Pay for Machines,” aiming to let AI agents buy and pay across cards, bank rails and stablecoins

Mastercard is laying groundwork for a world where AI agents, not people, initiate transactions, using a new payments concept called Agent Pay for Machines (AP4M). The initiative is intended to let machine-initiated services be requested and paid for automatically, with settlement routed across familiar payment channels such as cards and bank accounts, and also through stablecoin-based rails, according to early reporting on the announcement.

The proposal centers on “trusted” payment plumbing for AI-driven commerce, an issue Mastercard and others see as increasingly urgent as more software systems act on behalf of users and organizations. In that model, an AI agent would select a service, execute a purchase, and then complete settlement, ideally with payment acceptance and reconciliation that look similar to today’s commercial payments flow, even though the initiator is software rather than a cardholder.

Early details point to support for multiple payment types. Reporting describes AP4M as a service that can use cards, bank accounts, and stablecoins for settlement. Stablecoins are crypto tokens typically designed to track the value of a currency, often the US dollar, and they have become a common bridge for cross-border and on-chain payments. Mastercard’s stated direction suggests it wants AI agents to be able to pay in whatever format a recipient can accept, rather than forcing a single payments standard for machine commerce.

Partnership involvement is a notable feature of the rollout. Coverage says Mastercard is working with crypto and fintech ecosystem players such as Coinbase and Ripple, among others, to build what it frames as a credible environment for autonomous transactions. CoinDesk’s reporting similarly characterizes the company’s efforts as collaboration with firms including Coinbase and Stripe to build trusted payment systems tailored to AI-driven commerce.

The program’s broader ecosystem footprint also appears to be substantial, based on multiple third-party writeups. One recent report describes Agent Pay for Machines being launched with a roster of partners that includes a mix of legacy finance and crypto entities, with references to over 30 participants across fintech and crypto infrastructure. The breadth of the partner list matters because machine-to-machine payments need endpoint coverage, including payment service providers, settlement networks, and compliance operators that can handle transactions at scale.

Ripple’s own role is also consistent with this direction. While that company’s latest effort appears separate from Mastercard’s AP4M branding, reporting indicates Ripple has launched tools aimed at enabling AI agents to pay with XRP and RLUSD on its XRPL network. That kind of integration work reflects the parallel industry trend of “agentic payments,” where agent platforms and settlement networks align on how value moves and how accounts, confirmations, and returns are handled.

Still, many details remain unclear from the available public reporting. The Decrypt and Yahoo Finance summaries, as well as other secondary coverage, do not provide a full technical specification, pricing model, geographic availability, or timelines for when AP4M will be broadly deployable. It is also not yet clear what levels of authentication and risk controls Mastercard expects for fully autonomous purchases, especially in scenarios where an AI agent could place high volumes of transactions quickly.

For the payments industry, the competitive implications are straightforward: Mastercard is attempting to position itself as the interoperability layer for AI commerce, spanning both traditional rails and crypto-native settlement. If AP4M matures as described, it could help merchants and service providers accept machine-initiated payments without rebuilding their payment stacks from scratch. The next watch items are whether Mastercard names specific merchants or platforms piloting the system, how it handles compliance and chargebacks for automated transactions, and whether stablecoin settlement becomes a meaningful portion of the early adoption.