Mastercard launches Agent Pay for Machines, a payment protocol built for AI agents

Global card payments giant, Mastercard, has on Tuesday unveiled the Agent Pay for Machines (AP4M) protocol that allows AI agents to send and settle payments, including microtransactions costing less than a cent, across the card network’s global infrastructure.

More than 30 companies have signed on as early partners, a list that includes Stripe, Coinbase, Cloudflare, Polygon, OKX, and Ant International, according to a press release from the company.

Mastercard’s AP4M use case

The protocol is designed to help AI agents in purchasing services from other agents, autonomously and at high speed. Mastercard states that this is a distinct category from traditional point-of-sale or e-commerce payments, which are initiated by a person and processed one at a time.

AP4M handles transactions that are programmed, continuous, and executed between systems without a human having to supervise and “confirm.” For example, an entrepreneur could tell an AI agent to build a flower shop’s web presence, and that agent would independently purchase a domain, hosting, images, and checkout pages within a set budget, chaining multiple payments across providers from a single prompt.

“Machine payments can make it possible for services to be bought and sold among agents at fundamentally different scales than payments today,” Mastercard chief product officer Jorn Lambert said in the press release. He described these scales as “very high volumes, very small values, very fast and at extremely low latency.”

Built on blockchain permissions

A design choice that stands out is Mastercard choosing to record the permissions that humans grant the AI agents on public blockchains rather than in a private database. The initial chains selected are Polygon, Solana, and Base.

This helps with verification, as multiple parties can independently check if an agent is operating within the boundaries its human owner set, without relying on any company’s records.

Polygon confirmed the partnership on X, writing that it would “help establish common rules for agent-to-agent use cases and accelerate the adoption of agentic commerce, with always-on settlement powered by Polygon.”

Agents need payment rails that run at machine speed.

We’re excited to join @Mastercard’s Agent Pay for Machines, to help establish common rules for agent-to-agent use cases and accelerate the adoption of agentic commerce, with always-on settlement powered by Polygon. pic.twitter.com/WGp2MupktB

— Polygon | POL (@0xPolygon) June 10, 2026

Mastercard looking at future of AI

Mastercard CFO Lambert mentioned that this product is driven by future benefits and not immediate rewards for the company. “Am I expecting that this is going to be a huge revenue driver for Mastercard next year? No,” he told Fortune. He added that he sees AP4M becoming “a meaningful new addressable market” over the next five years.

This timeline fits the wider AI agent market regarding payments. Visa, Stripe, and Google have all released their own agent payment tools or standards over the past year; however, transaction volumes remain a small fraction of the commercial flow currently.

Lambert predicted that AI chatbots will eventually dictate and handle a meaningful share of online purchases, stating that “it’s just easier for consumers to do.” On machine-to-machine payments specifically, he was not as certain but said “too much is happening in that space” for the industry to fail entirely.

Mastercard has not announced a public launch date for AP4M beyond the initial partner phase.