A new report from digital asset market company Keyrock shows that the infrastructure needed for artificial intelligence (AI) programs to make online payments is advancing rapidly. Despite being a niche sector for now, leading technology, payments, and cryptocurrency firms are fast-tracking their efforts in this space. According to the Keyrock report, between May 2025 and April 2026, AI agents are projected to conduct over 176 million transactions on blockchain networks, with a total value exceeding $73 million.
Major players lead the race in machine payments
Industry giants such as Visa, Stripe, Google, and Coinbase are each developing solutions for automated machine-to-machine payments. For example, Visa processes about $14.5 trillion in annual transaction volume, highlighting that AI-driven payments are still tiny compared to traditional finance. However, the report emphasizes that the speed of ecosystem development and ongoing infrastructure investment are far more significant than transaction volume at this stage.
The general vision in this field centers on software that can independently purchase digital services without human intervention. For instance, an AI trading algorithm might buy data, cloud services, or analytics continuously throughout the day with no need for manual approval.
Forecasts show rapid expansion ahead
Gartner predicts that AI-mediated transactions could reach up to $15 trillion annually by 2028. McKinsey estimates that by 2030, retail machine commerce could hit a volume of $3 to $5 trillion. Keyrock suggests that this pace of growth may even outstrip the sharp rise witnessed in the stablecoin market. Additionally, progress on the infrastructure side indicates the industry is quickly moving past the experimental phase.
Coinbase’s x402 protocol is emerging as a key system in the sector. This solution lets AI applications make direct payments with $USDC on blockchain, without requiring them to set up accounts or subscriptions.
Mini-glossary: The x402 protocol, created by Coinbase, is a blockchain system that enables AI programs to pay for services instantly, without identity checks or manual approval.
Stripe is targeting the market through its Machine Payments Protocol (MPP), which operates on the Tempo blockchain. Google has launched AP2, a framework that authorizes spending on behalf of AI software. Visa, meanwhile, is adding new token-based authorization features to its card network specifically for AI-driven purchases.
Crypto-based infrastructures and stablecoins continue to gain traction as the backbone for machine-to-machine transfers, mainly due to their minimal fees.
Conventional rails not suited for low-value payments
The report finds that 76% of AI program payments fall below the standard 30-cent card processing fee, with the majority in the 1 to 10 cent range. These low, frequent transactions make traditional payment systems inefficient for automated machine activity. Conversely, moving funds with $USDC over blockchains like Base or Tempo cuts costs below even one cent.
Today, 98.6% of all machine-to-machine payments are completed using $USDC. Issued by US fintech firm Circle, $USDC is a stablecoin designed to always maintain a 1:1 peg with the US dollar. This strengthens Circle’s key position in the industry but also raises concerns about concentration risk since so much payment flow depends on a single issuer.
Lack of regulation is sector’s biggest roadblock
The report identifies regulatory uncertainty as a core obstacle for the sector’s growth. Major frameworks such as MiCA in Europe, the GENIUS Act in the US, and the EU’s AI Act are expected by mid-2026. Still, as of today, none of these laws directly address issues such as machine-to-machine transactions, responsibility, or digital intermediary identity.
The Keyrock report notes that while faster infrastructure construction signals the end of the machine payments trial phase, the lack of regulation creates risk and uncertainty across the industry.
