If you look at the first decade of Web3, most of the energy went into trading: tokens, leverage, speculation. Payments were the subplot everyone liked to talk about but rarely used in anger. Traditional rails were “good enough” for most people, stablecoins looked niche, and it was hard to argue convincingly that on-chain payment beat SEPA or card schemes for a human doing a normal transaction.

That picture is already breaking. Stablecoins processed an estimated $27-32 trillion of value in 2024, more than Visa and Mastercard combined, with several trillion tied directly to cross-border payments. On any given day, they support $20-30 billion of genuine on-chain payment flows, not just exchange churn. In parallel, AI agents are moving into production: they’re calling APIs, negotiating contracts and triggering transactions, often with no human in the loop. Web2 rails were never built for machines to pay machines, which is why we’re seeing protocols built for agent-to-agent micropayments, streaming and escrow that settle natively on-chain.

Two additional shifts matter. First, regulation is becoming a feature. With MiCA now live, the teams gaining traction are the ones treating compliance as part of the architecture:  regulated stablecoins, custody and settlement layers with built-in KYC/AML, and analytics designed for institutions rather than traders. Second, tokenized real-world assets (RWAs) are moving from theory to balance sheet use. Treasuries, credit and revenue-share products are being adopted because they improve liquidity and settlement, not because they are crypto.

Together, these trends point to programmable finance: agents don’t just need to pay; they need programmatic access to treasury, credit and revenue-share logic. Web3 is becoming the backend where that can run safely.