AI software is increasingly acting as an autonomous spender, using cryptocurrency to purchase data, compute power and access to AI models. Chainalysis-derived data shows that the x402 payments protocol on Coinbase’s Base network logged over 138 million transactions in the past year, the majority of which were valued at fractions of a cent.
Scale of agentic commerce
The x402 protocol, incubated by Coinbase, is designed for machine-to-machine payments. Analysis of its activity reveals that more than 87% of the recorded payments fell between one-hundredth of a cent and ten cents. In July alone, the protocol processed roughly 75 million payments amounting to about $24 million over a 30-day span. The rapid funding pattern—wallets created in Q2 2026 received additional capital within 30 days at an 89.3% rate, and the interval between initial funding and first payment shrank from 73 days in late 2025 to just 18 minutes in Q3 2026—suggests that agents are quickly becoming active blockchain participants.
Geographic split of spending
When the source of the wallets can be identified, U.S.-linked addresses dominate purchases of AI model inference, accounting for 88.1% of such payments. By contrast, Chinese-linked wallets contributed only 0.1% of inference-related spending but represented roughly 90% of payments for security and code-analysis services, much of it directed to the AgentLisa smart-contract auditing platform. Analysts note that Chinese developers often rely on open-weight models like DeepSeek and Qwen, reducing their need for metered inference services.
Stablecoins as the preferred medium
The prevalence of sub-cent payments highlights the suitability of stablecoins for this emerging use case. Traditional card networks cannot economically process transactions of such a small size, whereas stablecoins can settle fractions of a cent instantly. Industry voices, including Cloudflare’s chief strategy officer, argue that stablecoins may underpin the next wave of crypto users—potentially billions of AI agents—by providing a cost-effective medium for automated purchases.
Institutional backing and future outlook
Major payment processors such as Visa, Mastercard and Stripe have publicly supported the x402 protocol, and Coinbase now allows business customers to accept payments from AI agents. Yat Siu, co-founder of Animoca Brands, projects that up to 100 billion AI agents could be transacting online in the coming years, many using stablecoins. While the sheer number of transactions demonstrates growing adoption, the modest monetary value raises questions about the depth of genuine commercial demand.
Why it matters
The emergence of AI agents as independent economic actors signals a shift in how digital services are procured and paid for on blockchain networks. The dominance of U.S. wallets in inference spending points to a regional concentration of AI-driven commerce, while China’s focus on security services reflects divergent development strategies. Stablecoins’ ability to handle micro-transactions without prohibitive fees could unlock new business models, prompting traditional payment infrastructure to adapt or risk obsolescence. As autonomous software continues to scale its on-chain activity, regulators, businesses and developers will need to monitor this nascent market for signs of sustainable demand and potential systemic impacts.
