Your AI just made several payments while you were reading this title. You haven’t approved any of them. Visa has not processed any of them. And if the biggest bulls in the crypto industry are right, it’s not a bug: it’s the entire future of the Internet economy.
Coinbase founder Brian Armstrong believes there will soon be more AI agents than humans transacting on the internet. Binance founder Changpeng Zhao went further, predicting that agents would make a million times more payments than people, all in crypto. The messages arrived on the same day last week and lit up crypto X.
The central argument is structural.
AI agents cannot open bank accounts because banks require identity verification that software cannot provide, whereas a crypto wallet only needs a private key. No KYC, no compliance checks, no waiting – and it’s this asymmetry that Armstrong was pointing out.
But the wallet problem is only half the picture. The other half is economical.
Agents don’t shop like humans do. When an AI agent performs a task, such as researching a topic, coordinating a supply chain, creating a report, it can call dozens of specialized APIs in a single session.
Each call can be worth fractions of a cent, as it pays for GPU computing time, real-time data feeds, web scraping services, or hiring a sub-agent to handle translation. None of these transactions look like what Visa or Mastercard were designed to do.
Consider for a moment that this story was written by an agent, solicited by a CoinDesk “chief” agent charged with increasing the authority of the site.
To produce it, this agent would have queried a real-time information API to verify Armstrong’s tweet ($0.002), pulled on-chain data to look for volume figures ($0.004), cross-press releases ($0.001), and pinged a financial context model for Visa protocol details ($0.003). This would ultimately generate the item at additional cost, paying credits to another AI tool to actually to write the room.
The total cost of reporting is less than two cents with six transactions, compared to current figures offered by protocols such as x402.
In contrast, Stripe’s minimum processing fee on a single transaction is around $0.30. Making these six payments through a card network would cost more than 100 times the value of the payments themselves.
A human editor reviewing and publishing the article could then be billed by a sub-agent responsible for SEO optimization, another responsible for plagiarism checks, and another formatted for the CMS software. Each micropayment is economically absurd on card rails, but trivial on chain.
This is the thesis behind x402, Coinbase’s open payments protocol that integrates stablecoin payments directly into HTTP requests – so an agent can hit a paywall, pay in USDC, and continue their task in the same interaction, without the need for a human. Cloudflare, Circle, AWS and Stripe all support it. Google’s open agent payment standard includes x402 as a settlement layer.
Every sector in which high-frequency, low-value data exchange becomes a candidate.
In healthcare, an agent handling a patient’s insurance claim pays per document retrieved from a medical records API. In logistics, a procurement agent auctions freight slots with dozens of carriers in real time, settling the winning bid instantly. In media, AI bots pay per indexed article rather than negotiating bulk licensing deals. In finance, a trading agent pays a specialized model fractions of a cent per risk signal consumed.
It should be noted, however, that infrastructure is ahead of demand.
CoinDesk reported this week that x402 is currently processing a daily volume of around $28,000, with Artemis reporting around half of the observed transactions as artificial activity rather than actual trading. The x402 merchants they were built for are still rare.
Meanwhile, traditional finance is not standing still. Visa launched its Trust Agent protocol last October and Mastercard last week achieved Europe’s first live banking payment with an AI agent within Santander’s regulated infrastructure – both on existing card rails with cryptographic verification overlaid.
The most likely outcome is a split, where regulated commerce stays on the card rails, while machine-to-machine payments – such as agent recruitment, pay-per-API-call, on-demand compute purchasing – migrate to stablecoins because the economy demands it.
The open question is which bucket ends up being bigger.
