More than 138 million transactions moved through Coinbase-incubated x402 on Base by the end of Chainalysis’s 2026 reporting period. The firm’s North America adoption study found that autonomous software used the protocol to buy model inference, data, search, code analysis and other machine-readable services.
The distribution shows why stablecoins are being tested for machine commerce. More than 87% of the analyzed payments ranged from one-hundredth of a cent to 10 cents, values that are generally uneconomic for card networks. Chainalysis also found that the share of newly funded x402 wallets making a payment within one hour rose from 4% in the third quarter of 2025 to 67% in the third quarter of 2026.
Repeat funding improved as well. Among wallets first funded in the second quarter of 2026, 89.3% received a second funding within 30 days, compared with 70.6% for the third-quarter 2025 cohort. Chainalysis excluded wallets active on only one day from that retention comparison, an important methodological limit that reduces one-off experiments but does not establish whether every remaining wallet represents an independent commercial user.
Geography is highly concentrated. U.S.-linked wallets generated 88.1% of geolocated model-inference payments, while Chinese-linked wallets accounted for only 0.1% of that category and dominated the smaller security and code-analysis segment. Attribution depends on Chainalysis’s geolocation methods, and the visible transaction count says little about gross revenue because most payments are tiny.
TOKEN RECON ASSESSMENT
x402 has moved beyond a demonstration in transaction count and wallet behavior, but the economic proof remains incomplete. The decisive metrics are recurring spend per funded wallet, independent buyers, service-provider revenue and fraud or refund rates. A network can process enormous machine traffic while producing modest dollar demand, so transaction count should not be treated as token value capture.
Sources: Chainalysis research, CoinDesk.
