Agentic Commerce and the Economics of Machine-to-Machine Payments

We define agentic commerce as the delegation of commercial decisions and payments to artificial intelligence (AI) and provide statistics on prices and services using data from Coinbase’s x402 Protocol and Tempo & Stripe’s Machine Payments Protocol. Despite more than 234 million recorded transactions, activity across these protocols consists largely of pay-per-use micropayments for digital resources, while observed x402 usage is concentrated among a small number of services, many of which report few paying accounts. Agentic commerce can reduce search and transaction costs, but it also creates governance and accountability problems when AI agents make persistent errors, interact strategically, or are manipulated, potentially generating negative distributional and welfare effects. This review also discusses regulatory approaches and identifies priorities for AI governance and future research.