Artificial intelligence agents have completed more than 138 million blockchain transactions through Coinbase's x402 payment protocol on the Base network, according to new research published by blockchain analytics company Chainalysis on October 14, 2026. The findings highlight the rapid development of autonomous payment systems that let AI-powered software buy digital services, access information, and complete transactions without human intervention for every payment.
The research, published as part of Chainalysis' 2026 analysis of cryptocurrency adoption in North America, shows machine-to-machine payments moving beyond early experimentation toward more regular commercial activity. More than 87% of the transactions examined involved payments worth between $0.0001 and $0.10, demonstrating the importance of low-cost payment infrastructure for AI systems that frequently purchase small amounts of data, computing capacity, and other digital resources.
Although these transactions don't match conventional retail purchases, their growth signals a key development in agentic commerce: AI systems are becoming economic participants that can manage limited spending decisions and purchase services as part of automated workflows.
Chainalysis found that cumulative transactions using the x402 protocol on Base had exceeded 138 million by its 2026 reporting period. The figure represents a substantial increase from the approximately 1,500 transactions recorded in mid-2025, when the technology was still in its earliest stages of adoption.
Developers and businesses have driven the expansion by experimenting with autonomous software capable of making payments directly to online services. Instead of requiring a human user to manually authorize every transaction, an AI agent can use a funded digital wallet to purchase resources as needed to complete an assigned task.
For example, an AI system conducting financial research might need access to a premium market-data service. A software development agent might need a code review, while another assistant might need to buy more computing capacity or access a specialized AI model. Through compatible payment infrastructure, these transactions can occur automatically within the agent's workflow.
However, the headline figure requires an important distinction. The 138 million transactions represent the number of payments processed through x402 on Base, not the monetary value of those payments. They should not be interpreted as $138 million in revenue or as 138 million consumer shopping orders.
Chainalysis also identified substantial speculative activity during the protocol's early development, particularly involving cryptocurrency tokens. Consequently, transaction growth alone does not establish the scale of commercially productive AI activity.
Nevertheless, the research provides measurable evidence that autonomous payment infrastructure is attracting sustained interest.
Coinbase introduced the x402 protocol in May 2025 to enable automated payments directly through internet requests. Its name references HTTP status code 402, traditionally associated with the concept of payment being required to access an online resource. The technology allows software applications and AI agents to purchase services without relying on conventional subscription arrangements or repeated manual payment procedures.
When an agent requests access to a paid resource, the service can respond with payment instructions. The agent then evaluates the request, transfers the required amount through a supported blockchain payment mechanism, and provides confirmation to access the resource. Payments commonly use stablecoins, including USDC, which are designed to maintain a relatively stable value against a reference currency such as the US dollar.
The protocol supports multiple blockchain networks, though Base, developed by Coinbase, has emerged as a particularly active environment for x402 transactions. This architecture's economic significance lies in its ability to support very small payments at relatively low settlement costs.
Traditional payment cards are designed primarily for consumer and business transactions involving amounts large enough to justify processing fees and other operational expenses. Payments worth fractions of a cent are generally uneconomical through conventional card infrastructure. Blockchain-based settlement systems can make such transactions more practical, allowing digital services to charge per request rather than requiring customers to buy subscriptions or prepaid packages. This could support new commercial models in which AI systems purchase precisely the information or computing resources they need.
One of the most revealing findings concerns the value of individual transactions. Chainalysis examined x402 activity over the 16 months ending in September 2026 and found that more than 87% of payments fell between $0.0001 and $0.10. The most common transaction values were between one and ten cents. These figures indicate that much of the emerging agentic payment economy is based on micropayments rather than conventional retail transactions.
For developers, the ability to charge small amounts for individual services creates opportunities to commercialize digital resources that would otherwise be difficult to sell economically. For example, a financial information provider could charge an AI agent for each dataset requested. A software security company might charge for individual code inspections, while an AI infrastructure provider could charge per model query. This approach differs from the subscription-based pricing structures that dominate many software services.
Instead of committing to a monthly payment, autonomous systems could access resources on demand, with costs determined by actual usage. Such arrangements may become especially important as AI agents grow more sophisticated and coordinate multiple specialized services to complete complex tasks. For the broader digital economy, micropayment infrastructure could help create marketplaces where software applications automatically exchange services and information.
However, high transaction volumes should not be confused with high commercial revenue. The predominance of extremely small payments means that even millions of transactions may represent comparatively modest financial value.
Beyond transaction volumes, Chainalysis identified changes in how quickly AI-related wallets begin spending after receiving funds. In late 2025, the median interval between a wallet's initial funding and its first x402 payment was approximately 73 days. By the third quarter of 2026, that interval had fallen to just 18 minutes.
The finding suggests developers increasingly create and fund wallets specifically to run automated tasks, rather than for general experimentation or holding digital assets. The proportion of newly funded wallets making their first x402 payment within one hour also increased substantially. In the third quarter of 2025, only 4% of wallets made their first payment within an hour of receiving funds. By the third quarter of 2026, the figure had reached 67%.
This acceleration suggests a closer connection between wallet funding and immediate operational requirements. An AI agent may receive a defined spending allocation shortly before beginning a task and then use those funds to purchase the services necessary to complete it. Such behavior aligns with the development of automated payment workflows in which spending decisions are part of software execution.
Nevertheless, the data does not establish that every transaction was initiated independently by an AI model. Some activity may involve automated applications or scripts operating within predefined instructions.
Chainalysis also examined whether wallets continued receiving funds after their initial activity, providing another indicator of recurring usage. Among qualifying wallets first funded in the third quarter of 2025, 70.6% received additional funding within 30 days. For the corresponding group first funded in the second quarter of 2026, that proportion increased to 89.3%.
The analysis focused on wallets that had made payments on at least two separate days, excluding one-time participants from this particular comparison. This qualification is important because the figures measure repeat funding among an already active group rather than retention across all newly created wallets.
Even with that limitation, the increase suggests developers and organizations using x402 are becoming more willing to replenish wallets and keep automated payment systems running. Repeat funding can indicate that an agent is performing tasks valuable enough to justify ongoing expenditure.
For example, a business might allocate a recurring budget to an automated market intelligence assistant that purchases information from external data providers. Similarly, software development teams could fund agents that continuously access testing tools, security services or specialized computing resources. These use cases show how autonomous payments could become a routine part of business operations.
Chainalysis's findings show that the emerging market for autonomous payments is currently concentrated in digital services rather than physical consumer goods. One of the largest spending categories is AI model inference, where an agent pays to access another AI model.
This can occur when a relatively simple agent needs specialized capabilities to complete a task, such as analyzing complex information, generating code or interpreting technical documents. Other important categories include data and search APIs, financial intelligence, software code review, security analysis and tools used to develop or operate AI agents.
In effect, AI systems are increasingly purchasing services from other software systems. This development could create a digital economy in which agents act simultaneously as buyers, service providers and intermediaries.
A business-oriented AI assistant, for example, might buy market data from one provider, get analytical processing from another, and use a third service to generate a report. Each interaction could involve a separate automated payment, potentially completed within seconds.
The model gives providers of specialized digital services opportunities to monetize their capabilities through standardized payment infrastructure. At the same time, Chainalysis cautions that some early x402 transaction activity was driven by speculative cryptocurrency applications, including memecoin-related transactions. The emergence of more sustained information-service payments is an important development, but the overall ecosystem remains early-stage.
The research also identifies substantial geographic differences in the development of autonomous payment activity. According to Chainalysis, wallets attributed to the United States accounted for approximately 30% of geographically identifiable transactions between AI agents and services using x402.
The American share was considerably higher in categories associated with AI infrastructure. US-linked wallets represented 88.1% of identifiable transactions involving AI model inference and 41.8% of payments to agent frameworks and tool catalogs. They also accounted for 23.2% of transactions involving data and search APIs.
These figures reflect the prominent position of American companies in artificial intelligence development, blockchain infrastructure and digital payment technologies. Coinbase, which developed x402, and Circle, the issuer of USDC, are among the companies supporting the infrastructure used for autonomous payments.
The United States is also home to major AI model developers and a substantial ecosystem of businesses building agent-based software. China presents a different pattern. Chainalysis identified particularly strong Chinese-linked activity in security and code analysis services, suggesting the emerging agentic economy may develop different regional specializations.
However, geographic attribution in blockchain analysis has limitations. Wallet activity cannot always be reliably assigned to a particular country, and the available percentages refer only to transactions for which geographic identification was possible.
Although the Chainalysis findings primarily concern digital services, they have broader implications for e-commerce development. Agentic commerce describes a commercial environment in which AI systems can participate in purchasing decisions, product discovery and transaction execution. For consumers, this could eventually involve shopping assistants that compare products, identify suitable retailers and facilitate purchases according to predefined preferences.
For businesses, autonomous agents may increasingly purchase software services, market intelligence, logistics information and other resources required to support commercial operations.
The x402 protocol demonstrates one possible payment infrastructure for this emerging environment, particularly where transactions are frequent and individually small. However, conventional retail requirements differ from those of machine-to-machine micropayments. Purchasing physical products involves inventory availability, delivery addresses, taxation, consumer protection, fraud prevention, returns, and dispute resolution.
These considerations make it unlikely that a single payment protocol will address every aspect of agentic commerce. Instead, different payment technologies may serve complementary purposes. Blockchain-based stablecoin payments could become increasingly relevant for automated purchases of digital services, while established payment providers continue supporting consumer transactions that require additional authorization and protection mechanisms.
The growth of x402 should therefore be seen as evidence of expanding autonomous payment capabilities, not proof that blockchain transactions are replacing conventional e-commerce checkout systems.
Machine-to-machine payments could be particularly relevant for cross-border digital commerce. Businesses operating internationally often encounter payment processing costs, currency conversion expenses and administrative requirements when purchasing services from providers in other countries. For small digital transactions, these costs can become disproportionately high.
Low-cost blockchain payments could reduce some of these barriers by letting software applications settle transactions using compatible digital currencies and payment protocols. This may create opportunities for international providers of data, computing services and AI infrastructure to sell access to customers without requiring conventional billing relationships for every transaction.
An AI agent operating in Europe, for example, could theoretically purchase access to a specialized information service located in another region through an automated stablecoin payment. Such transactions could support more flexible international markets for digital resources.
Nevertheless, cross-border adoption will depend on several factors, including regulatory compliance, sanctions screening, cybersecurity, stablecoin availability and the legal treatment of digital asset payments.
Businesses will also need mechanisms to manage automated spending limits, transaction records, and financial accountability. Consequently, autonomous payment infrastructure will need appropriate governance and risk-management systems.
Chainalysis research shows that autonomous payment systems are developing more persistent usage patterns. The increase to more than 138 million cumulative x402 transactions, combined with faster initial spending and higher repeat funding among active wallets, suggests that developers are integrating automated payments into increasingly practical workflows.
However, the market remains relatively immature. Most transactions involve very small amounts, early adoption has been driven by speculative activity, and available research does not establish the total economic value generated by autonomous payments. Nevertheless, the findings indicate that a new form of digital commerce is beginning to take shape.
Rather than relying exclusively on human users to initiate purchases, software applications are becoming capable of acquiring resources and services as part of their normal operations. For e-commerce companies, payment providers, and technology developers, this creates opportunities to build infrastructure that supports transactions between people and autonomous systems.
The immediate commercial impact may be most visible in digital services, where AI agents can purchase information, computing resources and specialized capabilities without the logistical complexity of physical retail. Over time, however, the same underlying principle could influence broader commerce: software systems identifying commercial requirements, evaluating available suppliers and executing payments within defined limits. Chainalysis' findings suggest that this transition is already underway, even if the full economic potential of agentic commerce has yet to be demonstrated.
By continuing to use the site, you agree to the use of cookies. more information
The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.
