AI Agents Make 138M Crypto Payments as US Leads Spending

Autonomous artificial intelligence software isn't just generating text or creating images anymore. These software programs, often called AI agents, are now spending real money on digital services using cryptocurrency.

According to a report by blockchain analytics firm Chainalysis titled Inside North America's $3 trillion crypto economy: Agentic commerce accelerates, AI agents have completed more than 138 million transactions using the x402 payment protocol on Coinbase's Base blockchain.

The Rise of Machine-to-Machine Crypto Payments

AI agents use cryptocurrency to purchase data, raw computing power, and access to complex artificial intelligence models. They operate autonomously, executing tasks without needing direct human approval for every single purchase.

Much of this fast-growing activity relies on x402, an open payments protocol incubated by Coinbase. The system allows automated software to send micro-payments instantly across the blockchain.

"AI agents are becoming economic actors where every agent is effectively a new 'user' of the blockchain," explained Tejwani in the report.

United States Dominates Spending on AI Models

The research reveals a massive geographic split in how AI agents spend money online. Wallets linked to the United States accounted for 88.1% of all identified payments made to purchase access to AI models.

In contrast, wallets linked to China represented just 0.1% of spending on AI model inference.

Why is there such a massive gap between the two countries? Chainalysis points to differences in software tools and technical preferences.

"One possible explanation is that Chinese developers are more inclined to use open-weight models such as DeepSeek and Qwen, reducing their need to purchase metered inference services," said Eric Jardine, Head of Research at Chainalysis.

China Focuses on Automated Security Audits

While Chinese wallets spent very little on AI model access, they dominated a completely different corner of the market.

Chinese-linked wallets generated roughly 90% of all geographically identified payments for security and code-analysis services.

A primary driver of this activity was Agent Lisa, an AI-powered platform designed to audit smart contracts for code vulnerabilities and safety flaws.

However, Jardine cautioned against drawing broad conclusions about regional technological strength based on these payment flows alone.

"That's one possible interpretation, but we wouldn't go so far as to say open-source AI is starving Asia's on-chain inference market," Jardine noted.

Why Micro-Transactions Require Stablecoins

Traditional payment networks like credit cards are not designed for small machine-to-machine payments. Standard payment processors charge processing fees that make tiny purchases financially impractical.

Chainalysis found that over 87% of analyzed x402 payments were worth between one-hundredth of a cent ($0.0001) and 10 cents ($0.10).

• Traditional credit cards charge minimum processing fees that swallow small payments.
• Stablecoins on layer-2 blockchains like Base settle fractions of a cent instantly.
• AI agents can buy tiny fragments of data or computing cycles on demand as needed.

"Stablecoins are particularly well suited to the kinds of transactions we expect AI agents to make at first," said Stephanie Cohen, Cloudflare's Chief Strategy Officer.

Major Financial Giants Back Agentic Commerce

The underlying infrastructure supporting AI payments is gaining rapid institutional backing across both traditional finance and crypto.

Heavyweight payment networks—including Visa, Mastercard, and Stripe—have backed the x402 protocol.

Meanwhile, Coinbase has enabled its corporate customers to start accepting automated payments directly from AI agents.

Looking at the Numbers: Real Demand or Network Activity?

While transaction counts sound massive, the total dollar values remain relatively modest for now.

In July 2026, the x402 protocol processed approximately 75 million payments. However, those 75 million transactions totaled just $24 million over a 30-day period.

Researchers also tracked how quickly newly created wallets began making automated transactions:

• Among wallets first funded in Q2 2026, 89.3% received additional funds within 30 days, up from 70.6% in Q3 2025.
• The average time between a wallet receiving funds and making its first x402 payment fell from 73 days in late 2025 to just 18 minutes in Q3 2026.

These rapid speeds show that software setups are becoming far more streamlined. However, researchers noted that transaction volume alone does not prove genuine commercial demand.

A Future Filled With Billions of AI Users

Industry leaders believe that automated software agents could eventually dwarf the human user base on crypto networks.

"Over the next few years, I estimate up to 100 billion AI agents will be transacting on the internet and they will be using crypto," said Yat Siu, co-founder and chairman of Animoca Brands.

As software agents take on a bigger role in online commerce, stablecoins and fast layer-2 blockchains appear set to power this new automated economy.

Disclaimer: Past performance of digital assets or operational metrics does not guarantee future results. Digital asset markets carry risk, and readers should evaluate technology trends independently.