AI Agents Get Their Own Payment Systems
A new trend is emerging where fintech and crypto firms are building payment systems specifically for AI agents, often using stablecoins to bypass traditional card networks. This convergence of AI agents, stablecoins, and open banking is redefining the future of autonomous finance and operational roles within banks.
Payments giants Circle and Stripe are building the foundational rails for this economy. Stripe has funneled over $1.1 billion into stablecoin infrastructure, acquiring firm Bridge in 2025, while Circle has launched Arc, a dedicated blockchain for stablecoin payments, and is testing "nanopayments" for machine-to-machine commerce. This emerging market for AI-driven payments currently sits at around $50 million across 40,000 on-chain agents—a tiny fraction of the $46 trillion settled annually using stablecoins. However, the autonomous finance market is projected to grow by $13.54 billion between 2024 and 2029, with some analysts like Gartner predicting over 80% of finance functions will embed AI autonomy by 2030. The shift is driven by cost and speed, as traditional card networks are poorly suited for the high-frequency, low-value transactions that define an agentic economy. Stablecoin transactions can cost less than one cent and settle almost instantly, making automated micropayments economically viable for the first time. Incumbent networks are not standing still. Visa is developing "Visa Intelligent Commerce" and Mastercard has launched "Mastercard Agent Pay," both of which use secure network tokens that allow authorized AI agents to initiate payments without exposing underlying card details. Stripe is the first to support both these new agentic network tokens and Buy Now, Pay Later services like Klarna and Affirm for agent transactions. An entire technical architecture is being built to support this shift. An agent-based payment relies on three pillars: API-first processing, tokenized credentials to protect raw card data, and pre-authorized frameworks that set rules for what an agent can purchase and spend. New standards like the Coinbase-backed x402 protocol, which uses the HTTP 402 "Payment Required" code, are emerging to facilitate these automated transactions. The technology is still in its infancy and carries significant operational risks. In one testing incident, an AI agent accidentally transferred $450,000, highlighting the fragility of these early-stage systems. Furthermore, current stablecoin systems lack the robust fraud protection and dispute resolution frameworks that have been built into credit card networks over decades
