TL;DR — Key Takeaways
-
AI agents are becoming active digital buyers: More than 138 million x402 transactions on Coinbase’s Base blockchain highlight the rapid growth of AI-driven payments.
-
Micropayments are fueling agentic commerce: More than 87% of x402 payments are worth 10 cents or less, enabling AI agents to purchase computing power, data and other services.
-
AI agents are spending faster and more frequently: The median time between wallet funding and the first payment fell from 73 days to just 18 minutes.
AI agents have completed more than 138 million cryptocurrency transactions on Coinbase’s Base blockchain, according to new research from Chainalysis, indicating a major surge in the use of autonomous agents in purchasing online services.
The transactions were conducted using x402, a payment protocol that enables AI agents to spend cryptocurrency without requiring human approval for each purchase. The research found that U.S.-linked digital wallets accounted for 88.1% of geographically identified payments for AI model inference, compared with just 0.1% for Chinese-linked wallets.
AI agents are now more frequently given their own spending budgets to purchase computing resources, access AI models and obtain data needed to complete tasks. A key technology supporting this activity is stablecoins, cryptocurrencies typically pegged to the U.S. dollar. Developers can deposit stablecoins, including Circle’s USDC, into digital wallets controlled by AI agents. The agents then use these funds to purchase services as needed.
Unlike traditional credit card systems, blockchain payments can handle transactions worth only fractions of a cent. But even this capability is important because many AI services involve extremely small purchases. Chainalysis found that more than 87% of x402 payments were valued between $0.0001 and 10 cents.
For enterprises using AI agents, this payment model could enable automated workflows that require frequent purchases of specialized services without repeated human intervention.
Businesses Paid by AI Agents
Coinbase introduced x402 in May 2025 and subsequently partnered with Cloudflare to develop the protocol. The initiative has attracted major financial and tech companies. In July 2026, an x402 foundation was launched under the Linux Foundation with 40 members, including Google, Mastercard, Stripe and Visa.
The involvement of established payment providers suggests that agent-based commerce is moving beyond cryptocurrency companies toward broader commercial adoption.
Coinbase has also begun enabling business customers to receive payments directly from AI agents.
Chainalysis found that cumulative x402 transactions on Base grew from approximately 1,500 in mid-2025 to more than 138 million by the third quarter of 2026. But transaction volume does not necessarily translate into substantial revenue. During one 30-day period in July, the protocol handled approximately 75 million payments totaling $24 million.
AI Agents Are Spending Faster
Beyond transaction growth, Chainalysis identified changes in how AI agents use their digital wallets. By the third quarter of 2026, the reported time between a wallet receiving its initial funding and making its first payment had dropped to 18 minutes, compared with 73 days in late 2025. About two-thirds of newly funded wallets now complete a payment within an hour.
Developers are also replenishing these wallets more frequently. Among qualifying wallets initially funded in the second quarter of 2026, 89.3% received additional deposits within 30 days. The figure for wallets first funded in the third quarter of 2025 was 70.6%.
In other words, some developers are moving beyond isolated tests toward sustained AI agent operations.
The geographic distribution of transactions reveals contrasting approaches to AI development. While U.S.-linked wallets dominate payments for AI inference, Chinese-linked wallets accounted for approximately 90% of geographically identified transactions involving security and code-analysis services. Much of that Chinese activity involved AgentLisa, an AI-based smart contract auditing platform.
Eric Jardine, Chainalysis’ head of research, suggested that Chinese developers’ use of open-weight models, including DeepSeek and Qwen, could help explain their limited spending on paid inference services.
