Autonomous agent economy faces infrastructure gaps
Visa and Artemis say the autonomous agent economy faces infrastructure bottlenecks that block broader commercial adoption. Traditional card rails built for low-frequency human spending struggle with high-frequency AI micropayments. Agents need near-zero fees and faster settlement before agentic commerce can scale into the mainstream. A joint report explains why payments rails lag agent capabilities and how hybrid card-stablecoin paths could follow.
Key Takeaways
- Visa and Artemis link slow agent commerce adoption to infrastructure bottlenecks, not a lack of AI capability.
- Card networks designed for infrequent human checkouts struggle with high-frequency agent micropayments.
- Agents need near-zero fees and quicker settlement for commercial-scale machine payments.
- Stablecoins and cards may converge in hybrid workflows rather than compete head-on.
- Early protocols such as Coinbase’s x402 already show rising machine-payment volume.
What did Visa and Artemis find about agent payments?
Payments giant Visa and investment thesis platform Artemis released a joint report stating that infrastructure gaps are limiting mainstream use of autonomous AI agents in commerce. According to Cointelegraph’s reporting, incumbent global card rails are under pressure from high-frequency micropayments that agents initiate.
The report argues traditional cards were built for human commerce with relatively low transaction frequency. That design is a poor fit for AI agents that may trigger many small payments in rapid succession.
For readers tracking fintech and crypto payment shifts, more coverage lives in our Fintech & Crypto Alerts section.
Why does the autonomous agent economy face friction now?
Visa and Artemis say AI agents crossed a key capability threshold in mid-2025. Agents can discover unfamiliar APIs, evaluate prices, and decide on autonomous payments—raising demand for machine-native rails.
Commercial viability, they argue, depends on infrastructure with near-zero fees and faster settlement. Without that, agentic micropayments remain hard to scale even as software ability improves. In short, the autonomous agent economy faces a plumbing problem more than a model problem.
Australian exchange Swyftx separately estimated that AI-enabled microbusinesses could add about $262 billion in stablecoin volume by 2033 through AI-native payments, assuming roughly 33% adoption. That outlook underscores why settlement speed and cost now sit at the center of the debate.
Are stablecoins and cards competing or converging?
Visa and Artemis describe a convergence path: cards for proxy purchases inside existing merchant networks, stablecoins for machine-native micropayments, and hybrid flows that use both in one workflow. A single machine-payments framework could support stablecoin and traditional card transactions, they said.
Coinbase’s x402 payment protocol is cited as an early adoption signal. Since launching in May 2025, it processed $15 million in adjusted volume across more than 109 million adjusted transactions. Volume accelerated sharply in October 2025, when monthly transactions rose from 40,000 to 3.8 million, with 38 million processed that month alone.
Tempo’s Machine Payment Protocol spans onchain crypto and fiat via shared payment tokens, while Visa’s Card Specification SDK aims to extend agent commerce into card rails. Visa’s crypto unit and Stripe-backed Tempo both launched AI payment tools in March, including Visa support for same-day agent payments.
Closing fee and settlement gaps—and stitching card plus stablecoin rails together—will likely decide how fast agent commerce moves from pilots to everyday commercial use.