The world of payments for artificial intelligence is beginning to clearly divide into two segments, and, as recent research shows, stablecoins play a key role here in micropayments. In their joint analysis, experts from Visa and the blockchain analytics company Artemis concluded that traditional cards and digital assets will occupy different but complementary niches.
The authors identify two fundamentally different categories of AI agent transactions. The first is purchases made on behalf of a human: booking tickets, managing subscriptions, ordering goods. These operations are essentially no different from regular online commerce, and traditional payment cards are ideally suited for them. The second category is frequent microtransactions worth less than one dollar that occur between programs. An AI agent may pay for each individual request to a cloud service, for obtaining data, for computing power, or for using software tools.
This is precisely where traditional card infrastructure fails. Fixed fees make cards completely unprofitable for such small payments. At the same time, in the Base and Solana networks, transaction fees amount to fractions of a cent. This makes stablecoins an ideal tool for settlements between AI agents.
Protocols and Real Volumes
The report examines two key open protocols: x402, whose main activity is concentrated in the Base, Solana, and Polygon networks, and the Machine Payments Protocol (MPP), which supports settlements both through blockchains and through traditional payment systems. According to Artemis data as of April 21, after filtering out fake and test transactions, approximately $15 million passed through x402 across 109.6 million transactions. For MPP, the figures are more modest — roughly $25,000 and 115,000 transfers, respectively.
It is important to note that the original unfiltered data looks much more impressive: $135.7 million and 178.3 million operations through x402. The 15-fold difference further emphasizes how important it is to critically evaluate the loud numbers in this field. The market for AI agent payments is still in its formative stage, and a significant portion of the activity consists of tests and experiments.
The Main Problem — Trust
Despite technological readiness, the authors point to a key unresolved issue — trust. Who will be responsible for an erroneous or unauthorized purchase: the owner of the AI agent, the platform operator, the model developer, or the seller? Existing payment reversal rules (chargebacks) are designed for human actions. There is simply no mechanism for disputing thousands of automatic transactions per hour. Additional risk comes from malicious instructions capable of redirecting a payment or forcing an AI agent to make an unplanned purchase.
In my opinion, we are witnessing the emergence of a hybrid model where cards will remain for authorization and large purchases on behalf of humans, while stablecoins will occupy the niche of a "circulatory system" for settlements between programs. However, until a clear system of responsibility and transaction dispute mechanisms is established, the mass adoption of AI agent payments will remain in question. This is not a technical problem, but a legal and ethical one that the industry must solve.