AI agent payments still tiny but infrastructure is racing ahead
A new report from crypto firm Keyrock suggests that artificial intelligence agents are starting to use blockchain-based payment systems more frequently, even if the total value involved remains small today. Keyrock, which focuses on crypto trading and investment, estimated that AI agents settled over $73 million across roughly 176 million transactions on blockchain rails between May 2025 and April 2026. To put that in perspective, Visa alone processes around $14.5 trillion annually. So the current numbers are basically negligible compared to traditional finance.
But here’s the thing: the report argues that the real story isn’t about the dollar volumes yet. It’s about how quickly the underlying infrastructure is forming. Big global firms including Coinbase, Stripe, Google, and Visa have all launched competing systems specifically designed for machine-to-machine payments. That’s a lot of major players moving fast.
The basic idea behind agentic payments is simple: software is increasingly buying digital services on its own, without humans managing subscriptions or authorizing every transaction. An AI trading agent might purchase market data, cloud computing time, or AI analysis in tiny amounts throughout the day, automatically. No human needs to click approve each time.
Forecasts suggest explosive growth potential
Those kinds of use cases are driving some pretty ambitious predictions. Gartner thinks AI agents could intermediate $15 trillion in purchases by 2028. McKinsey estimates retail agentic commerce could hit $3 trillion to $5 trillion by 2030. According to Keyrock, those growth rates could actually surpass what stablecoins experienced during their breakout years. And the pace of infrastructure buildout already suggests the market is moving past pure experimentation.
Coinbase’s x402 protocol is one of the leading crypto-native systems right now. It lets AI agents pay directly with USDC for services like blockchain analytics or cloud infrastructure, without needing to create accounts or manage subscriptions. Stripe launched its own competing framework called Machine Payments Protocol (MPP) on its Tempo blockchain. Google introduced AP2, which focuses on delegated spending authorization for AI agents. Visa has also extended its card network with tokenized credentials designed for AI-driven commerce.
Why crypto rails make economic sense for agents
Crypto rails and stablecoins look like the preferred settlement option here, and the economics help explain why. According to the report, about 76% of agent transactions fall below the 30 cent fixed-fee floor that’s common in card payments. Most payments range between one and 10 cents. Traditional payment rails just aren’t practical for automated software agents buying tiny bits of data, AI inference, or API access. Meanwhile, stablecoin settlement on some blockchains like Base and Tempo costs fractions of a cent.
Currently, 98.6% of machine payments settle in USDC, the stablecoin issued by Circle. That solidifies Circle’s position in crypto payments, but it also creates concentration risk. The whole system depends heavily on a single issuer.
Regulatory uncertainty looms
Regulation could become a constraint too. Europe’s MiCA framework, the U.S. GENIUS Act, and the EU AI Act are all expected to take effect around mid-2026. But none of them directly address autonomous machine-to-machine transactions or questions about liability and agent identity. That’s a big hole, and it might slow things down if not clarified soon.
