AI agents now pay each other on blockchain rails — do different countries specialize in different slices of that spending?

This explores where the wallets that pay for AI agent services on blockchain are located, and whether different countries lead in different kinds of spending. The corpus has one direct data source here, and it shows real regional specialization. Chainalysis traced the location of wallets making x402 payments (x402 is a protocol that lets agents pay each other small amounts per request). American wallets made 88.1% of payments for AI model inference, meaning agents paying to run a model, and 41.8% of payments to agent frameworks. Chinese wallets clustered in a different category: security services Which countries dominate agent-to-agent AI payments on chain?. So the split isn't simply that one country is ahead. Wallets in different countries are buying different layers of the agent stack.

The split matters more once you know the spending has become routine. A companion Chainalysis finding shows that the median time from funding an agent wallet to its first payment fell from 73 days to 18 minutes, and that the share of wallets topped up again rose from 71% to 89% Are AI agent wallets becoming genuinely operational instead of experimental?. Wallets that get funded and spent almost at once, then refilled, look like operating budgets rather than speculation. That suggests the regional pattern reflects real infrastructure purchasing, not a passing memecoin crowd.

A useful lateral comparison is Anthropic's Economic Index, which found that Claude usage tracks GDP per capita. Wealthier countries use it for many kinds of work, while poorer countries focus mostly on coding Does AI adoption follow wealth and mature over time?. The x402 data may show the same pattern in payment form: where AI use is broad and mature, spending covers core inference and orchestration, and elsewhere it narrows to particular niches. That is an inference across two datasets, not something either study tests directly.

The gap you should know about: the corpus has a single geolocated source on this question. It doesn't break down Europe, other parts of Asia, or the Global South. It also doesn't explain why security services cluster in China, or whether the American lead in inference simply reflects where the big model providers and their payment endpoints are based. Read the Chainalysis note as a first snapshot, not a settled map.