More than 138 million transactions have passed through the x402 crypto-payment protocol on the Base blockchain, according to new research examining the growth of AI-driven commerce.
Chainalysis published the findings on 14 October as part of its 2026 North America crypto-adoption report. The activity includes payments for AI-model access, code analysis, search tools, data feeds and other digital services that software can purchase while completing a task.
The headline count demonstrates substantial protocol activity, but it should not be interpreted as 138 million individual customers or purchases made by fully independent AI systems.
Key Takeaways
- Chainalysis counted more than 138 million x402 transactions on Base.
- More than 87% of payments in its extended sample were worth between $0.0001 and $0.10.
- US-linked wallets accounted for 88.1% of geographically identified payments for AI-model inference.
- Services receiving payments included AI models, data APIs, code-review tools and software for building agents.
- The transaction count measures blockchain activity—not the number of users, AI agents or profitable businesses.
How x402 Payments Work
The protocol uses the HTTP 402 “Payment Required” response to allow a website or online service to request payment before providing a digital resource.
Compatible software can receive the request, send a blockchain payment and continue the task after payment is confirmed. A developer can therefore fund a wallet and allow an AI agent to pay for approved services without manually authorising every small transaction.
These transactions differ from ordinary online purchases because many involve machine-readable resources rather than physical goods. Examples include an API response, access to an AI model, a search result or an automated software-security check.
UPay’s explanation of how AI bots can use payment cards examines another route for automated payments. Card-based and on-chain systems can support similar workflows, but they use different authorisation, settlement and dispute processes.
Why Most Payments Are So Small
More than 87% of the transactions analysed in Chainalysis’s extended payment-size sample fell between one hundredth of a cent and ten cents.
Payments at this scale can make sense when software purchases one API request or a small piece of data at a time. A human user would be unlikely to complete a checkout process for every fraction-of-a-cent purchase, while software can execute repeated transactions automatically.
Understanding how crypto payments work helps explain why blockchain-based settlement is being considered for these machine-to-machine interactions. Digital assets can move between compatible wallets without requiring a person to enter card details for every request.
However, a large number of micropayments does not automatically represent a large amount of revenue. The reported distribution covers a wide range, and multiplying the total transaction count by an assumed average would not produce a reliable market value.
What the 138 Million Figure Does Not Show
A transaction is an action recorded on a blockchain, not necessarily a separate customer.
One funded wallet can make thousands of requests. A developer can operate several wallets, while several agents may share the same wallet. Promotional campaigns, token issuance and automated testing can also generate substantial activity without representing sustained commercial demand.
The count is therefore useful evidence of x402 protocol usage, but it cannot independently establish:
- The number of distinct AI agents
- The number of human users behind them
- Total merchant revenue
- Profitability of the services receiving payments
- The share of activity that continued over time
Chainalysis also notes that early x402 activity included short-lived interest involving speculative tokens. More durable categories now include information services, development tools and AI-model access.
US Wallets Dominate Identified AI Spending
US-linked wallets represented 88.1% of geographically identified payments for AI-model inference. Their shares were lower in other categories, including agent frameworks, tool catalogues, data services and search APIs.
These percentages apply only to wallets that could be geographically associated. They should not be treated as a complete map of every user or developer participating in the protocol.
The research also concerns x402 activity on Base. It does not cover every automated crypto payment across all blockchains, nor payments made by AI systems through bank transfers or card networks.
Controls Remain Important
Allowing software to make payments introduces operational questions about budgets, permitted recipients and transaction limits.
An automated system should not have unrestricted access to funds merely because each individual payment is small. Repeated micropayments can accumulate, and faulty instructions or compromised credentials could trigger unintended spending.
Crypto-payment gateways can help businesses manage parts of the process, including conversion and compliance. UPay’s guide to the role of crypto payment gateways explains how these services connect blockchain payments with broader merchant systems.
Conclusion
The x402 figures show that automated crypto payments have moved beyond isolated demonstrations. Software is already using on-chain transactions to purchase data, model access and digital tools at a scale that would be difficult to manage through manual checkout processes.
The 138 million total nevertheless measures protocol transactions rather than autonomous customers or economic value. Continued wallet funding, repeat usage and verifiable merchant revenue will provide stronger evidence of whether agent-driven payments are developing into a durable commercial market.
