Autonomous software programs are increasingly handling real-world money, with more than 138 million cryptocurrency payments recorded on Coinbase’s Base layer-2 network during the past twelve months. The activity, tracked through the x402 protocol, is dominated by micro-transactions that fall between a hundredth of a cent and ten cents, a price range that stablecoins can accommodate but traditional card systems cannot.
Scale of agentic commerce
Chainalysis, a blockchain analytics firm, reported that cumulative payments on the x402 bridge grew from a few thousand in mid-2025 to well over one hundred million by the third quarter of 2026. Developers are allocating budgets to AI agents so they can purchase computing power, live data feeds, and coding utilities on demand. While the sheer volume is striking, the majority of transfers are tiny, reflecting the low-value, high-frequency nature of the use case.
Geographic concentration
The data show a pronounced U.S. presence, with wallets linked to American entities responsible for 88.1% of traceable payments related to AI model inference. By contrast, Chinese-linked wallets contributed only 0.1% of that category, though they represented most spending on security and code-analysis services. Analysts caution that the figures capture only identifiable activity on the x402 protocol and do not reflect the full global market.
Infrastructure and industry response
The rapid rise of agentic payments coincides with a broader rollout of supporting services. Major card networks such as Visa and Mastercard have announced plans to accommodate AI-driven transactions, while Stripe has added similar capabilities. Coinbase has introduced tools that let agents trade, transact, and enables business customers to accept such payments. Google also unveiled a dedicated protocol for agent payments, underscoring the convergence of traditional finance and emerging AI commerce.
Skepticism about the numbers
Earlier research from a competing analytics firm suggested that a portion of the recorded activity could stem from scripted operations or self-dealing rather than truly autonomous agents. This highlights the difficulty of distinguishing genuine AI-initiated commerce from automated scripts that merely simulate agent behavior. Nonetheless, the overall trend points toward a growing ecosystem where software, rather than individuals, becomes a primary participant in financial flows.
Why it matters
The emergence of AI agents as regular participants in the crypto economy could reshape demand patterns for blockchain networks. Frequent, low-value payments may drive the adoption of more efficient scaling solutions and encourage stablecoin usage for micro-transactions. Moreover, the involvement of established payment processors signals a willingness to integrate AI-centric financial models, potentially accelerating mainstream acceptance of decentralized commerce. As the technology matures, the line between human and machine-driven transactions will blur, prompting regulators and market participants to reconsider how financial activity is monitored and taxed.
