AI Agents Make 138 Million in Crypto Payments as US Dominates AI Model Costs
According to new research by Chainalysis, US-linked wallets account for more than 88% of tracked AI model payments, compared with China's 0.1% share.
Artificial intelligence (AI) agents don't just talk to each other anymore. They pay for services, buy data and even access tokens for AI models, with more than 138 million transactions recorded via the x402 protocol on Coinbase's Base Blockchain, according to Chainalysis' $3 Trillion Crypto Economy Inside North America: Agent Commerce Accelerates report.
The United States dominates this emerging market. The blockchain analytics firm found that wallets associated with the United States accounted for 88.1 percent of the geographic payments identified for inference by the AI model, compared to only 0.1 percent for wallets associated with China.
"One possible explanation is that Chinese developers are more inclined to use open weight models like DeepSeek and Qwen, reducing their need to purchase metered inference services," Eric Jardine, head of research at Chainalysis, told CoinDesk.
The findings, published Wednesday, come as payment giants Visa, Mastercard, Stripe and others have backed the x402 protocol, while Coinbase has begun allowing business customers to accept payments from AI agents.
Yat Siu, co-founder and president of Animoca Brands, recently told CoinDesk that in the next few years, he estimates that up to 100 billion AI agents will be transacting on the Internet and using cryptography. And the next billion crypto users may be AI agents paying with stablecoins. Stablecoins are particularly well-suited for the types of transactions we expect AI agents to perform, Cloudflare Chief Strategy Officer Stephanie Cohen told CoinDesk first.
Unlike traditional card payments, stablecoins can handle transactions worth a fraction of a cent, allowing AI agents to purchase data, computing power and other services without human intervention. New figures from Chainalysis provide evidence of growing use of these payment systems, although the number of transactions alone does not determine how much business demand is true.
Most of the activity is done through x402, a payment protocol incubated by Coinbase that allows artificial intelligence agents to automatically pay for digital services using cryptography.
More than 87 percent of x402 payments analyzed by Chainalysis were between one-hundredth of a cent and 10 cents, making it uneconomical to process through traditional credit card networks.
Small amounts remain. In July, x402 processed approximately 75 million payments worth $24 million in 30 days. Previous research has also raised questions about artificial activity in the network.
Among wallets first invested in Q2 2026, 89.3% received additional funding within 30 days, compared to 70.6% for wallets first invested in Q3 2025. This analysis excluded wallets that only paid out on one day.
The time between receiving funds by a wallet and making the first x40 payment has also been drastically reduced, from an average of 73 days in late 2025 to just 18 minutes in Q3 2026.
Meanwhile, China showed a different pattern of activity. Associated with ChinaQif, the funds account for approximately 90% of geo-identified payments for security and code analysis services, most of which go towards AgentLisa, an AI-based smart contract auditing platform.
He cautioned against interpreting the gap as evidence that open-source AI is undermining Asia's payments market.
"That's one possible interpretation, but we wouldn't go so far as to say open-source AI is going to starve the chain inference market in Asia," he said.
These figures measure identifiable transactions through x402 rather than the entire global AI payments market.
"Artificial intelligence agents are becoming economic actors where each agent is virtually a new 'user' of the blockchain," Tejavani said in the report.
At the moment, transactions are small, but Chainalysis' findings show that independent software is starting to develop its own spending habits, and the US is leading the AI paid services market.
Diversified RWA stablecoins maintain a 5-7% return on real credit as crypto capital drops to 4%. GENIUS increases performance off-chain. TAM increases to $4 billion in 3 years.