Crypto-enabled AI agents drive a maturing ecosystem, report finds

Artificial intelligence agents handling settlements in crypto have evolved from a speculative idea to a measurable part of the payments landscape over the past year. A collaborative study by crypto investment firm Keyrock, in partnership with Coinbase and the Tempo blockchain, shows machine-to-machine settlements totaling $73 million across 176 million transactions from May last year through April 2026.

Ben Harvey, a researcher with Keyrock, framed the development as a maturation of the field: “In the past 12 months, machine-to-machine payments have gone from concept to a developed ecosystem.” The report adds that incumbents have already deployed more than $8 billion in acquisitions to secure their foothold in what is becoming a new payment stack. This points to a marketplace where automation and programmable finance are increasingly central to everyday crypto activity.

Key takeaways

• AI agents settled $73 million across 176 million transactions between May 2025 and April 2026, according to Keyrock in collaboration with Coinbase and Tempo.
• By the end of Q1 this year, more than 104,000 AI agents were registered across 15+ directories, with the average transaction size around $0.31.
• Nearly all settlements—about 98%—were conducted in Circle’s USDC, highlighting the centrality of a single stablecoin issuer in the nascent AI-agent economy and raising concerns about systemic risk if reserve management or regulatory status changes.
• Industry signals point to rapid acceleration: Circle’s leadership has projected billions of AI agents operating with stablecoins in the coming years, and surveys show strong user appetite for AI-managed trading and portfolio decisions.

Rapid expansion of AI agent settlements

The scale of activity captured in the Keyrock report underscores a shift from experimental pilots to a functioning settlement layer for machine-driven commerce. The collaboration with Coinbase and Tempo aimed to quantify how far the ecosystem has come and where it might be headed as AI agents autonomously initiate and settle transactions across networks.

Harvey notes that the sub-dollar nature of many AI-agent payments exposes the inefficiency of traditional rails. He pointed out that a typical fixed processing fee—often around 30 cents per transaction—renders micro-payments uneconomical on conventional networks. The study emphasizes that stablecoins have become the practical settlement layer for sub-dollar machine payments, enabling cost-effective, near-instant settlements that would be impractical with legacy rails.

USDC’s central role and the risk it entails

Remarkably, the vast majority of AI-agent settlements—about 98%—used USDC, according to the report. While this confirms USDC as the de facto settlement currency for machine commerce, it also centralizes risk around Circle’s reserves, regulatory standing, and operational reliability. Harvey describes this reliance as both a validation and a vulnerability: “If Circle faces a regulatory challenge, a de-peg event, or even sustained downtime, the agent economy has no fallback.”

These observations contribute to a broader conversation about resilience in crypto payment ecosystems. As automation scales, the ecosystem’s dependence on a single stablecoin issuer could become a focal point for regulatory scrutiny and risk management strategies among startups, exchanges, and developers building AI-driven financial primitives.

Signals of acceleration and user appetite

The trend lines extend beyond settlement volumes. Circle’s leadership has publicly forecast a future in which billions of AI agents operate with stablecoins, representing a substantial expansion of automated, programmable payments. At the same time, user sentiment toward AI-managed finance remains cautiously positive. A CoinGecko survey of 2,632 crypto users found broad comfort with AI involvement in trading, with 87% willing to let AI agents manage at least 10% of their crypto portfolios.

Beyond these polls, the ecosystem is already seeing related product activity. Exodus recently launched an AI-agent-focused stablecoin on Solana, reflecting ongoing experimentation with specialized rails and instruments designed to support autonomous finance. Such developments illustrate a broader push to embed AI capabilities into the fabric of decentralized finance, from settlement layers to on-chain applications and liquidity provision.

What this means for users, developers, and investors

The emergence of AI-agent settlement raises practical considerations for users and builders. For users, the promise is faster, cheaper sub-dollar payments that unlock new use cases—micro-tokens for API calls, service payments, or on-demand access to data streams. For developers, the data suggests a growing demand for robust, interoperable agent frameworks that can operate across networks and wallets while maintaining security and auditability. Investors will likely watch for how the ecosystem handles concentration risks around stablecoins, especially in times of macro or regulatory stress, and which protocols diversify their settlement rails to avoid single points of failure.

As adoption accelerates, observers should monitor regulatory clarity around stablecoins and the resilience of crypto payment rails under stress. The next few quarters will reveal whether the AI-agent economy can sustain its early momentum or whether diversification of settlement assets and governance will become a prerequisite for scalable, permissioned machine commerce.

Readers should stay attentive to regulatory developments affecting stablecoins, the evolution of settlement-layer standards for AI agents, and the emergence of new tools designed to broaden anti-fragility in machine-to-machine payments.