On June 10, Mastercard launched something that could reshape how software buys and pays for things. Agent Pay for Machines — AP4M — is a framework letting AI agents authorize, coordinate, and settle transactions across Mastercard's global payments network. No checkout screen, no human tap-to-approve for every purchase.
The mechanics are straightforward: a user sets up an AI agent, grants it a spending budget and parameters, and those permissions are logged on public blockchain networks — Polygon, Solana, and Base — rather than locked in a private corporate database. Any party in a transaction can verify the agent is acting within its allowed scope. The result is always-on commerce where software transacts with software around the clock.
More than 30 companies signed on for the launch: Coinbase, RippleX, Stripe, Solana Foundation, Adyen, Cloudflare, OKX, MoonPay, Anchorage Digital, Polygon, and others. Settlement works via traditional payment cards, banking rails, or stablecoins — USDC, Ripple's RLUSD, and PayPal's PYUSD are all approved instruments.
Mastercard's Chief Product Officer Jorn Lambert was direct about the business case: "Do I think it'll be a meaningful new addressable market for us over the next five years? I think so." The logic is clear — every micro-transaction an AI agent makes to buy data, API access, or compute time is a potential fee. If agentic commerce scales, Mastercard wants to be the rails it runs on.
Visa, Coinbase, and Stripe have all built competing tools in the same space. Mastercard's edge is reach: acceptance at more than 90 million merchant locations worldwide, available immediately. The bet is that when AI agents go shopping, they'll use the same infrastructure as every other card payment.
