A TRM Labs report estimates that most x402 protocol payment volume is not from AI agents, finding $52.7 million in total settled value across 198.9 million settlements processed by known x402 facilitators. After exclusions, TRM estimated that between 0.6% and 7.5% of the remaining commerce by value appeared attributable to AI-agent-driven payments. The report also noted that facilitator-broadcast payment averages were under $1 and that USDC comprised 99.6% of the settled value.
TRM Labs examined $52.7 million in settled value across 198.9 million settlements processed by known x402 facilitators on Base, Solana, and Polygon since May 2025. The dataset covered facilitator-broadcast payments and identified settlement-level activity across those networks. TRM separated preliminary settlement totals from subsequent filtering steps to isolate likely commerce. The report quantified payment flows before and after exclusions.
TRM excluded self-payments, bulk flows from one or two payers, and sellers with fewer than 10 buyers as part of its filtering criteria, which reduced the dataset to $25.62 million in likely commerce. After those exclusions, TRM estimated that between 0.6% and 7.5% of the remaining commerce by value appeared attributable to AI-agent-driven payments. Facilitator-broadcast payment averages were under $1. USDC accounted for 99.6% of the settled value, equal to $52.47 million.
TRM called for more accurate registration, counterparty reputation information, and monitoring for large numbers of small payments. The report includes these recommendations.
TRM Labs observed temporal variation in x402 protocol payment activity across Base, Solana, and Polygon. In late 2025, the report identified activity that included meme-token minting and payments to one AI-analysis service. In early 2026 a single payment contract concentrated volume. Around midyear, the report identified a return of AI-service payments routed through an agent-payment router.
The report also documents several ecosystem initiatives referenced alongside on-chain activity. Binance’s August Agent OS launch included an x402 payment layer. Coinbase’s Base accelerator targeted agent and payments startups with a $1 million program. In May, Amazon announced AgentCore Payments with Coinbase and Stripe to enable agents to pay for online services with stablecoins.
The report pairs observed temporal shifts in x402 payments with contemporaneous ecosystem developments related to agent payments. These events are presented as part of the report’s documented activity timeline.
TRM Labs documented on-chain agent registries that allow individuals to declare ownership of an agent address, and noted that such declarations are voluntary and currently not widely utilized. The report described a modeling assumption that treats an address which explores multiple services and products as a true agent and an address that repeats the same price as behaving more like a script. TRM acknowledged that this modeling choice may understate the presence of single-purpose agents that repeatedly pay one service. These observations were presented alongside the dataset and filtering steps used to isolate likely commerce.
TRM recommended more accurate public registration of agents, the inclusion of counterparty reputation information in registries, and enhanced monitoring to detect cases involving large numbers of small payments. The report framed these recommendations as responses to observed gaps in on-chain identification and to the limits of the applied modeling approach. TRM linked the recommended measures to the voluntary nature of current registry declarations.
The report presents these observations, modeling caveats, and recommended actions within its findings. It records the voluntary use of public registration and the specific steps TRM proposes to improve attribution. The text pairs the identified limitations with targeted registration and monitoring measures.
