AI Agents Drive Crypto Payments as U.S. Leads Spending on AI Models

Highlights

AI agents have made more than 138 million transactions through Coinbase’s x402 protocol, paying in cryptocurrency for data, computing power, and AI services. U.S.-linked wallets accounted for 88.1% of geographically identified payments for AI model inference, while Chinese-linked wallets represented roughly 90% of identified payments for security and code-analysis services. More than 87% of analyzed x402 payments were worth between one-hundredth of a cent and 10 cents, highlighting stablecoins’ potential for micropayments. However, transaction totals alone do not establish that the activity reflects sustained commercial demand. The data describes payments identifiable through x402, not the entire global market for AI-related payments.

Sentiment Analysis

• Overall sentiment: cautiously positive. The reported transaction volume and increasing use of AI agents to pay for digital services suggest that autonomous software is beginning to participate in on-chain commerce.
• Positive indicators: More than 138 million x402 transactions were recorded, and the share of recently funded wallets receiving additional funds within 30 days was 89.3% for wallets first funded in the second quarter of 2026. The shorter interval between funding and a first payment also suggests faster activation.
• Neutral and cautionary factors: More than 87% of analyzed payments were worth between one-hundredth of a cent and 10 cents. Small payments demonstrate a possible technical advantage for stablecoins but do not, by themselves, prove meaningful or recurring business demand. The data also covers x402 transactions rather than the wider market.
• Regional contrast: U.S.-linked wallets led identified AI-model inference payments, while Chinese-linked wallets were prominent in security and code-analysis services. The report presents possible explanations without treating them as definitive.
• Visual sentiment assessment: Positive but qualified, reflecting evidence of adoption alongside unresolved questions about scale and commercial significance.

Article Text

Artificial intelligence agents are beginning to do more than exchange information: they can pay for digital services using cryptocurrency. According to a Chainalysis report, AI agents have made more than 138 million transactions through Coinbase’s x402 protocol on the Base blockchain. The reported purchases include data, computing power, and access to AI-model inference services. The activity offers an early view of software acting as a customer in cryptocurrency-based markets.

The geographic distribution differs by service. U.S.-linked wallets accounted for 88.1% of geographically identified payments for AI model inference, while Chinese-linked wallets accounted for 0.1% of those payments. Eric Jardine, Chainalysis’ head of research, told CoinDesk that one possible explanation is that developers in China may be more inclined to use open-weight models such as DeepSeek and Qwen, which could reduce their need to buy metered inference services. This is presented as a possible interpretation, not a confirmed explanation for the difference.

The reported pattern is different for security and code-analysis services. Chinese-linked wallets accounted for roughly 90% of geographically identified payments in those categories, with much of the activity directed toward AgentLisa, an AI-powered smart-contract auditing platform. Jardine cautioned against concluding from this regional difference that open-source AI is undermining Asia’s payments market. He said the findings do not justify claiming that open-source AI is starving Asia’s on-chain inference market.

x402 is a payments protocol incubated by Coinbase that enables AI agents to pay automatically for digital services using cryptocurrency. Its small transaction sizes are central to the reported use case. More than 87% of the x402 payments analyzed by Chainalysis were worth between one-hundredth of a cent and 10 cents. Such amounts are generally too small to process economically through traditional credit-card networks, whereas stablecoins can support payments valued at fractions of a cent.

These micropayments could allow software agents to purchase resources without a person initiating each transaction. Stephanie Cohen, Cloudflare’s chief strategy officer, told CoinDesk that stablecoins are particularly suited to the kinds of transactions AI agents are expected to make initially. Yat Siu, co-founder and chairman of Animoca Brands, told CoinDesk that he estimates up to 100 billion AI agents could be transacting on the internet over the next few years and using crypto. These statements describe expectations and estimates, rather than measured totals in the Chainalysis data.

The report also points to changes in wallet activity. Among wallets first funded in the second quarter of 2026, 89.3% received additional funding within 30 days, compared with 70.6% for wallets first funded in the Q3 2025. The analysis excluded wallets that made payments on only one day. The time between a wallet receiving funds and making its first x402 payment also decreased, from an average of 73 days in late 2025 to 18 minutes in Q3 2026. These measures indicate more rapid activity among the wallets covered, but do not independently establish the purpose or long-term value of each payment.

Transaction amounts remain modest in aggregate as well. In July, x402 processed roughly 75 million payments worth just $24 million over 30 days. Earlier research raised questions about whether some network activity was artificial. Consequently, a high count of tiny payments should not be treated as conclusive proof of broad, durable commercial demand. Chainalysis’ figures describe transactions identifiable through x402, not all AI-related payments worldwide.

The findings arrive as Visa, Mastercard, Stripe, and other payments companies have backed the x402 protocol, while Coinbase has begun allowing business customers to accept payments from AI agents. The report characterizes agents as emerging economic actors, with each agent potentially functioning as a new blockchain user. For now, the available figures show activity concentrated in small payments and a regional split between paid model inference and security services. They suggest that autonomous software is developing payment behaviors, while leaving the overall size and commercial sustainability of the market uncertain.

Key Insights Table