Artificial intelligence (AI) agents have completed more than 138 million transactions through the x402 protocol on Coinbase's Base blockchain, according to a new Chainalysis report. The findings show how autonomous software increasingly uses cryptocurrency to purchase data, computing services, and access to AI models without human intervention.
The United States leads payments for AI model services, accounting for 88.1% of geographically identified transactions. Meanwhile, China accounts for just 0.1% of that activity.
Chainalysis released the findings Wednesday in its report, Inside North America's USD3 Trillion Crypto Economy: Agentic Commerce Accelerates. The analysis comes as major payment companies explore systems that allow AI agents to make direct digital payments.
AI Agents Expand x402 Payments as Stablecoin Use Grows
The x402 protocol, which Coinbase incubated, allows AI agents to pay automatically for online services using cryptocurrency. Unlike traditional payment systems, it supports extremely small transfers that would otherwise face high processing costs.
According to Chainalysis, more than 87% of the x402 payments it examined ranged from USD 0.0001 to USD 0.10. These amounts fall below the levels that traditional credit card networks can process economically.
Meanwhile, stablecoins provide a way for autonomous programs to purchase digital resources without relying on conventional card transactions. The payments can cover data access, computing power, and AI model services.
Stephanie Cohen, Cloudflare's chief strategy officer, identified stablecoins as suitable for the types of payments AI agents could initially make.
Interest from established financial companies has also increased. Visa, Mastercard, and Stripe have backed x402, while Coinbase has started allowing business customers to accept payments from AI agents.
Still, transaction volume provides only part of the picture. In July, x402 processed approximately 75 million payments worth USD 24 million over 30 days. Earlier research also raised concerns about artificial activity.
US Wallets Dominate AI Payments as China Takes Another Path
Chainalysis found a major geographic divide in payments for AI model inference, which refers to running AI models to generate responses.
US-linked wallets accounted for 88.1% of geographically identified inference payments. Chinese-linked wallets represented just 0.1%, showing different activity patterns across the two markets.
Eric Jardine, Chainalysis' head of research, offered one possible explanation. Chinese developers may rely more heavily on open-weight AI models, including DeepSeek and Qwen, reducing demand for paid inference services. However, Jardine cautioned against treating the difference as evidence that open-source AI was weakening Asia's on-chain inference market.
China showed stronger participation in another category. Chinese-linked wallets accounted for approximately 90% of geographically identified payments for security and code-analysis services. Much of that activity involved AgentLisa, an AI-powered smart contract auditing platform.
The figures cover identifiable x402 transactions rather than the entire global market for AI payments. As a result, they do not establish overall regional spending across every payment network.
Read More: NVIDIA Pledges USD 250 Million for UK AI Safety Research as Agentic AI Risks Grow
Chainalysis Finds Faster Funding and Spending Among AI Wallets
Beyond geographic differences, Chainalysis identified changes in how quickly wallets receive and spend cryptocurrency. Among wallets first funded in the second quarter of 2026, 89.3% received additional funding within 30 days. That compared with 70.6% for wallets first funded in the third quarter of 2025.
The researchers excluded wallets that made payments on only one day. They also found that the average delay between initial funding and the first x402 payment declined sharply.
That interval fell from 73 days in late 2025 to just 18 minutes during the third quarter of 2026.
Yat Siu, Animoca Brands' co-founder and chairman, estimated that up to 100 billion AI agents could transact online over the coming years. He expects cryptocurrency to support those transactions.
Meanwhile, Tejwani described AI agents in the Chainalysis report as emerging economic participants, with each agent functioning as another blockchain user. The report's transaction figures track growing payment activity, although they cannot independently determine how much comes from genuine commercial demand.
Conclusion
Chainalysis recorded more than 138 million x402 transactions on Coinbase Base, with US-linked wallets dominating paid AI inference activity. Meanwhile, China led identified payments for security services. Faster wallet funding and widespread micropayments show changing transaction patterns, although the findings do not establish the full scale of genuine commercial demand.