The State of AI Agent Payments 2026 report found that 90.8% of sampled x402 payments were worth less than one cent. BeInCrypto Intelligence published the report in collaboration with AWS, Cloudflare, Chainalysis and Zoomex. The study tracked 6.4 million transactions over 35 days, with payments totaling $119,947. The report found that much of the payment activity concerned API access, computing capacity and automated requests for data.
The report's assessment gave micropayments 60.7 points on a 100-point scale, compared with 12.0 points for institutional financial workflows. The report found that 14 of 19 reviewed AI tools operated under predefined spending or trading restrictions. The report put AI-labeled interfaces at less than 0.55% of trading volume on Hyperliquid. The report found that Zoomex required people to confirm orders prepared through Zoomex workflows. The report found that six of 12 institutional workflows could start or settle payments within authorized limits. Whether businesses will entrust substantial amounts of money to agents remains uncertain.
THORWallet founder and CEO Marcel Robert Harmann expects agents to conduct more trading over the next 12 months. Platforms including Hyperliquid already provide APIs that support automated execution, but controlling access to funds remains a challenge. Harmann expects payments for data and computing to remain the largest category by transaction count, while payment volume likely grows as agents conduct more actual trading. Harmann sees threshold signatures as a way to reduce the risks of unrestricted wallet access. Harmann said an agent could hold one key share while an independent signer holds another and checks compliance with policy. Such arrangements could permit agents to initiate larger transfers without allowing agents to move money independently. Harmann expects regulators to introduce automated monitoring as transaction counts increase. Harmann argues that people cannot inspect every machine payment individually and that monitoring must match transaction speeds while people retain final authority.
ZeruAI founder and CEO Ashutosh Sahoo does not expect a substantial move beyond micropayments until financial systems are built specifically for AI. Sahoo argues that existing banking infrastructure, compliance arrangements and liability models were created for people and institutions rather than autonomous economic actors. Agents currently tend to operate under fixed spending limits, trading mandates and emergency shutdown mechanisms. Sahoo considers those arrangements supervised automation rather than independent financial activity. Sahoo argues that the turning point will come when financial systems are designed around agents' capabilities instead of adapting agents to infrastructure built for people. Sahoo's proposed approach could embed spending permissions and risk requirements in protocols rather than rely mainly on checks after payments occur. That approach raises questions about who sets the rules and who has authority to alter the rules. Sahoo also worries that financial systems designed by machines could eventually exceed human understanding, even when the systems' decisions have coherent logic.
HumanWrite Technologies founder and CEO Tamara Frater expects growth to vary because purchasing computing capacity is simpler than making investment decisions. Frater expects the next 12 months may involve more experimentation, but Frater does not expect an immediate transition from sub-dollar payments to widespread autonomous financial decisions. Frater warns that an agent could obey a spending limit while following misleading instructions or paying an incorrect recipient. Frater argues that spending caps limit exposure without ensuring that recipients, instructions or underlying information are correct. The report recorded two incidents involving a combined $281,200 sent to attacker-controlled wallets before recoveries. Those incidents illustrate why spending limits alone cannot determine which parties an agent should trust. Frater expects financial institutions and regulators to deploy AI to detect suspicious activity as automated payments increase. Frater wants faster oversight without weakening the ability to identify the people responsible for decisions.
The executives' views highlight the distinction between executing payments and earning trust to make financial decisions. Transaction counts provide a weak measure of adoption for investors. Sustained commercial payment value and businesses' willingness to reduce human intervention provide a better test. The initial phase of agent payments demonstrates that machines can make on-chain payments worth pennies. The next phase must demonstrate whether companies will entrust agents with substantial funds.
AI Agent Economy Payments Report: 6.4M Transactions Reveal Micropayment Dominance
BeInCryptoShare
Summary
A 2026 AI agent payments report tracked 6.4 million transactions over 35 days, totaling $119,947. Micropayments under one cent made up 90.8% of the volume, mostly for API access and computing. Micropayment efficiency scored 60.7, while institutional workflows scored 12.0. AI tools operated under spending caps, and AI-labeled interfaces represented less than 0.55% of Hyperliquid volume. Executives from THORWallet, ZeruAI, and HumanWrite discussed trust and infrastructure challenges. BTC dominance remains key in the daily market report.
Source:Show original
Disclaimer: The information on this page may have been obtained from third parties and does not necessarily reflect the views or opinions of KuCoin. This content is provided for general informational purposes only, without any representation or warranty of any kind, nor shall it be construed as financial or investment advice. KuCoin shall not be liable for any errors or omissions, or for any outcomes resulting from the use of this information.
Investments in digital assets can be risky. Please carefully evaluate the risks of a product and your risk tolerance based on your own financial circumstances. For more information, please refer to our Terms of Use and Risk Disclosure.