The number of virtual asset transactions executed by artificial intelligence (AI) agents has surged past 138 million, with the U.S. taking the lead in the AI model inference service payment market. This trend highlights the expansion of an automated payment ecosystem where AI agents purchase data and computing resources without human intervention.
On Wednesday, blockchain analytics firm Chainalysis reported that transactions using the x402 payment protocol on Coinbase’s Ethereum Layer 2 network, Base, have exceeded 138 million.
Developed by Coinbase, the x402 protocol allows AI agents to automatically acquire data, computing resources, and AI model inference services.
The majority of these transactions were micro-payments. Chainalysis found that over 87% of x402 payments ranged from 1 to 10 cents. This demonstrates that ultra-small transactions, which are typically cost-prohibitive on traditional credit card networks due to processing fees, are now feasible through virtual asset payment systems.
On a global scale, the U.S. dominated AI model inference service payments. Of the geographically identifiable transactions, U.S.-linked wallets accounted for a staggering 88.1%. In stark contrast, China-linked wallets represented a mere 0.1%.
Eric Jardine, Chainalysis’s head of research, posited that Chinese developers’ preference for open-source, weight-based AI models like DeepSeek and Qwen might explain the relatively lower demand for paid inference services in China.
The speed of AI agents’ payment activities is also ramping up significantly. The average time between a wallet’s initial funding and its first x402 payment has plummeted from 73 days at the end of 2025 to just 18 minutes in the third quarter of this year.
In related news, major payment giants including Visa, Mastercard, and Stripe have thrown their support behind the x402 protocol. Coinbase has also rolled out features enabling its corporate clients to accept payments from AI agents, further solidifying the protocol’s position in the evolving landscape of AI-driven transactions.
