Chainalysis: AI Agents Made 138 Million x402 Payments on Base

AI agents have made more than 138 million payments through x402 on Base, the blockchain analytics firm Chainalysis counted in a new report on North America's crypto economy. The headline number is not the interesting part. Over 87% of those payments were worth between a hundredth of a cent and a dime, a range where card networks stop being economical.

Chainalysis calls this agentic commerce: AI agents using crypto to pay autonomously. It grew quickly over the period the report covers, with the United States as the nexus.

The x402 findings sit inside a wider set of figures. The report puts North America's crypto economy at $3.02 trillion in the 12 months ending 1 July 2026, down 1.9% from the period before. The US accounted for $2.78 trillion of that, down 2.2%. Canada grew 2.3%, to about $250 billion.

Over 87% of x402 Payments Are Under a Dime

Most of the traffic is machines buying small things. The most common payment size falls between one cent and ten cents.

That matters because the amounts are too small to process economically on traditional card networks, which is the gap stablecoin rails are being built to fill. Circle described the split it expects as "card rails for consumer-prompted agent payments, stablecoin rails for machine-to-machine settlements that are too small for card economics."

Institutional interest shows up in the same report. Will Su, BlackRock's head of digital assets research, said the firm is watching agentic commerce closely "with AI agents potentially creating meaningful demand for frequent, low-value payments around the clock."

Because x402 is so new, Chainalysis extended its review window to the 16 months ending September 2026 for this part of the analysis.

US Wallets Took 88.1% of Inference Payments

The geographic split is the loudest finding. American wallets account for 88.1% of the AI model inference transactions Chainalysis could geolocate, 41.8% of payments to agent frameworks and tool catalogs, and 23.2% of payments to data and search APIs.

Chinese-attributed wallets look like a different customer. They took 46.7% of market and financial intelligence payments and 35% of payments to settlement rails, but just 0.1% of model inference. Almost all of their security and code analysis spending went to AgentLISA, a service that audits Solidity smart contracts using a language model.

Across agent-to-services transactions that Chainalysis can place on a map, the US holds a 30% share.

Base is where the volume sits, though x402 activity is being tracked on other networks too, including the XRP Ledger.

New Wallets Now Spend Within Minutes of Funding

The behaviour of new wallets changed more than the geography did. In Q3 2025, 4% of wallets made their first x402 payment within an hour of being funded. By Q3 2026 that was 67%, so roughly two in three new wallets spend almost immediately.

The average wait between a wallet receiving funds and making its first payment fell from 73 days in late 2025 to 18 minutes in Q3 2026. Wallets also came back for more. The share that received a second funding within 30 days of the first rose from 70.6% for wallets first funded in Q3 2025 to 89.3% for those first funded in Q2 2026.

Chainalysis tried to screen out tourists. It counted only wallets that sent on two or more separate days, to exclude one-and-done experimenters.

Hoolie Tejwani, head of Coinbase Ventures, said in the report that "the building blocks set in 2025 are turning into products with real users, liquidity, and markets in 2026."

What Chainalysis Does Not Measure

Two limits are worth carrying with the numbers. The figures measure identifiable transactions through x402, not the entire global market for AI payments. And the 88.1% inference figure rests on Chainalysis's own geolocation method, applied the way it treats other on-chain activity, so it describes where identified wallets cluster, not every payment made anywhere.

The 138 million is a cumulative count of Base transactions at the point the report was compiled, not a daily rate.

What happens next is measurable. If the share of wallets that top up a second time keeps climbing, and the time to a first payment keeps falling, the network is adding repeat users rather than one-off experiments. The next cut of this data will show which way it went.