FinanceShareShare this articleCopy linkX (Twitter)LinkedInFacebookEmailCrypto rails are becoming the default payment layer for AI agents, report says

A recent report by Keyrock indicates that blockchain-based stablecoins are increasingly becoming the preferred payment layer for AI agents, as conventional card systems struggle with micropayment transactions.

By Krisztian Sandor|Edited by Jamie Crawley May 24, 2026, 1:00 p.m. 2 min readMake preferred on

Robots (Unsplash/Sumaid pal Singh Bakshi/Modified by CoinDesk)

Key Points:

  • According to Keyrock, AI agents executed transactions exceeding $73 million across 176 million blockchain transactions in the past year, though this is a small portion of the overall payments market.
  • Major companies like Coinbase, Stripe, Google, and Visa are developing competing infrastructures for machine-to-machine payments as AI software increasingly makes autonomous purchases for data and digital services.
  • Currently, nearly all payments made by agents are processed in USDC, underscoring Circle's rising role in crypto transactions and the associated risks of relying on a single stablecoin issuer.

The market for AI agents spending money autonomously remains relatively small, yet prominent technology and payment firms are actively constructing the necessary infrastructure, as highlighted in a new report by Keyrock.

Keyrock's analysis estimates that from May 2025 to April 2026, AI agents completed transactions valued at over $73 million through approximately 176 million blockchain transactions.

This amount is modest when compared to traditional finance, where Visa alone handles about $14.5 trillion annually. However, the report emphasizes the importance of the rapid development of infrastructure rather than just the sheer dollar amount.

Global brands like Coinbase (COIN), Stripe, Google (GOOG), and Visa (V) are launching competitive systems for machine-to-machine payments.

The concept behind agentic payments is that software is increasingly capable of autonomously acquiring digital services without human intervention. For instance, an AI trading agent might continuously buy market data, cloud computing resources, or AI-generated insights incrementally throughout the day without needing a human to approve each payment.

This potential has led to optimistic forecasts regarding the growth of the agentic payment market. Gartner predicts that AI agents could facilitate transactions worth $15 trillion by 2028, while McKinsey estimates retail agentic commerce could reach between $3 trillion and $5 trillion by 2030, according to Keyrock's report.

Such projections suggest growth rates that may surpass those seen during stablecoins' initial rapid expansion, although the report notes that the swift pace of infrastructure deployment indicates the market is advancing beyond mere experimentation.

Among the leading crypto-native systems is Coinbase's x402 protocol, which enables AI agents to make payments directly in USDC for services like blockchain analytics or cloud infrastructure without needing to set up accounts or subscriptions.

Stripe has introduced a competing framework known as the Machine Payments Protocol (MPP) with its Tempo blockchain, while Google has developed AP2, a system designed for delegated spending authorizations for AI agents. Additionally, Visa is enhancing its card network with tokenized credentials aimed at AI-enabled commerce.

As crypto rails and stablecoins become the leading settlement layer, the economic rationale becomes clear.

About 76% of agent transactions fall below the typical 30-cent fixed fee associated with card payments, as reported. Most transactions are between one and ten cents, rendering traditional payment systems impractical for automated software agents making purchases for data, AI inference, or API access. In contrast, stablecoin settlements on various blockchains such as Base and Tempo incur costs of only a fraction of a cent.

Currently, 98.6% of machine payments are settled using USDC, the stablecoin issued by Circle (CRCL). This reinforces Circle's significant role in crypto payments but also raises concerns about the risks associated with dependency on a single issuer.

Future regulations may pose challenges to growth. The MiCA regulation in Europe, the U.S. GENIUS Act, and the EU AI Act are all projected to be implemented around mid-2026, but none of these initiatives directly address the complexities of autonomous machine-to-machine transactions or related issues regarding liability and agent identity, as noted in the report.

Artificial IntelligenceStablecoinsCoinbaseCircleStripeGoogle