Payment giant Visa, together with blockchain analytics company Artemis, conducted a large-scale study on the future of financial transactions between AI agents. The conclusion is clear: traditional cards and stablecoins will occupy clearly separated niches, and for micropayments, cryptocurrencies have no real competitors.
Analysts divided all AI agent operations into two fundamentally different classes. The first is purchases made on behalf of a human: booking tickets, managing subscriptions, ordering goods. Essentially, this is the same e-commerce, but with an automated trigger. Here, Visa or Mastercard cards are quite appropriate.
The second category is frequent transactions for amounts under one dollar that programs make among themselves. This includes paying for individual API requests, accessing computing power, or purchasing data. It is here, in my estimation, that the real breakthrough for stablecoins lies.
Why Cards Lose the Battle for Microtransactions
The fixed fees of card networks make every transaction under $1 economically unviable. At the same time, on the Base and Solana networks, transaction fees amount to fractions of a cent. Visa and Artemis highlighted two open protocols for such settlements: x402 (active on Base, Solana, and Polygon) and the Machine Payment Protocol (MPP), which supports both blockchains and traditional systems.
As of April 21, after cleaning data from fake and test transactions, approximately $15 million passed through x402 across 109.6 million transactions. MPP's figures are more modest: roughly $25,000 and 115,000 transfers. For comparison, raw unfiltered data showed $135.7 million and 178.3 million operations through x402 — a 15-fold difference, confirming the importance of verifying on-chain activity.
Hybrid Model as a Compromise
The study's authors suggest that the future lies in a hybrid model: cards will be used for authorization and large purchases on behalf of humans, while stablecoins will be used for instant settlements between programs. Both tools can be applied at different stages of executing a single task.
However, the key unresolved issue is trust. Who bears responsibility for an erroneous transaction made by an AI agent? The owner, the platform operator, the model developer, or the seller? Existing chargeback rules are designed for human actions, not for thousands of automated operations per hour. An additional risk is malicious instructions capable of redirecting a payment.
My comment: The market for AI agent payments is in its infancy, and stablecoins have a unique advantage here. Traditional fiat systems are simply incapable of efficiently processing millions of microtransactions worth cents. However, without solving the responsibility issue and creating mechanisms for disputing automatic payments, mass adoption will remain in question. Visa and Alchemy have already launched the AgentCard service with user-defined limits, but this is only the first step toward a fully autonomous economy.