In May 2026, at the Consensus conference in Miami, Richard Widmann, Head of Web3 Strategy at Google Cloud, and May Zabaneh, Senior Vice President of Crypto at PayPal, sent a clear message: The next generation of internet commerce, powered by AI agents, will run on crypto payment rails. This statement isn’t just business rhetoric—it’s rooted in the structural barriers AI agents face within traditional financial systems and the inherent technical compatibility of crypto infrastructure with autonomous machines. As millions of AI agents begin to handle complex economic tasks, they require a payment system—not designed for humans, but built to serve software.
Why Traditional Bank Accounts Fundamentally Cannot Serve AI Agents
AI agents aren’t unable to open bank accounts due to immature technology, but because the foundational architecture of traditional finance is incompatible with autonomous software. At Consensus, Widmann stated bluntly, "AI agents cannot open bank accounts. It’s not just difficult—it’s fundamentally impossible." The structural reasons span both technical and regulatory domains: Technically, traditional payment APIs and network authentication processes depend on human identity and interaction points, lacking programmable interfaces for autonomous agents. On the regulatory side, opening a bank account requires KYC (Know Your Customer) compliance, but AI agents are neither legal entities nor natural persons, making identity verification impossible. In contrast, crypto wallets are generated by private keys, requiring no account approval, making them inherently suited as settlement tools for machines. Widmann summarized this distinction: Crypto payments "are an excellent machine-readable payment interface."
How Much AI Agent Traffic Is Already Appearing on Merchant Platforms
The migration of commercial activity to AI agents depends first on merchant readiness. PayPal recently released survey data covering a broad range of merchants: Currently, 95% of merchant websites have seen AI agent traffic, but only about 20% offer machine-readable product catalogs. This data reveals a structural gap between supply and demand—AI agents are actively entering commercial scenarios, while most merchants still present products for human consumers. Zabaneh noted that this transformation mirrors the early shift from physical stores to online shops. Merchants must quickly display products in agent-readable formats or risk missing the next wave of commercial innovation. Compared to the offline-to-online transition, the penetration of AI agent commerce may be faster, as demand-side adoption is already underway.
How Crypto Payment Infrastructure Adapts to AI-Native Commerce
Crypto payment rails are architecturally primed to support autonomous AI transactions through four key attributes. First, no identity verification is required; agents sign transactions directly with private keys, bypassing account opening hurdles. Second, real-time final settlement—blockchain transactions are typically confirmed irreversibly within seconds, avoiding the "IOU-style" delays of credit card networks in banking systems. Third, programmability—smart contracts can predefine spending rules, limits, and automated execution logic, providing the exact financial management model needed for AI agents to operate autonomously "by the rules." Fourth, globalization and instant settlement—AI agents executing cross-border tasks across jurisdictions don’t rely on lengthy, multi-country banking networks. Zabaneh described PYUSD as "a very natural programmable payment layer," emphasizing stablecoins’ innate suitability for globalization, AI-native experiences, and tokenized assets.
How Multi-Party Custody Addresses AI Agent Fund Security
AI agents’ independent payment capabilities demand robust security mechanisms to prevent misuse of funds. Widmann clearly identified multi-party custody as the core solution for AI agent fund management. He explained that AI agents shouldn’t control the full private key but only hold one of two or three key shards. The logic is simple: Agents can initiate payment requests but cannot unilaterally transfer funds without signatures from other authorized parties. In consumer scenarios, this allows users or authorized entities to monitor and intercept agent payments in real time; in revenue scenarios, agent earnings must be allocated via the custody framework. Google has expanded its Cloud KMS platform into crypto custody, aiming to deploy multi-party custody mechanisms in cloud infrastructure for enterprise-grade fund security in large-scale AI agent deployments.
Why Stablecoins Are the Settlement Choice for AI Agent Commerce
Market data from 2026 has already confirmed stablecoins’ dominance in AI agent payments. According to Circle, over the past nine months, AI agents completed 140 million payments totaling $43 million, with 98.6% settled in USDC and an average transaction size of just $0.31. This means over 98% of autonomous machine economic activity has chosen stablecoins as the settlement tool, not traditional payment methods. The technological foundation behind this trend is evolving rapidly: In May 2026, Circle launched a USDC nano-payment system for AI agents, lowering the minimum transaction to $0.000001, eliminating user-side gas fees, and supporting 11 major blockchains. The system uses off-chain micro-payment aggregation and batch on-chain settlement, diluting gas costs to negligible levels and removing economic barriers for high-frequency, small-value payments. Experts predict that by 2030, AI agent-driven autonomous commercial transactions could reach $3–5 trillion, with stablecoins’ strategic value as the settlement layer continuing to grow.
How Open Standard Protocols Drive Industry-Wide Collaboration and Transformation
Scaling AI agent payments requires an interoperable, open protocol layer—not isolated, closed systems. Google has taken a key step by launching the Agentic Payments Protocol (AP2), now donated to the FIDO Foundation. The protocol currently boasts over 120 partners, including PayPal. Widmann likened this move to donating the x402 native payment standard to the Linux Foundation, emphasizing that "open dialogue and open standards are the foundation of infrastructure." Leading industry players are signaling collaboration: AWS has integrated the x402 payment protocol into its AI infrastructure via Amazon Bedrock AgentCore Payments, enabling AI agents to automatically pay for services in USDC. In May 2026, Exodus launched XO Cash, a stablecoin designed for AI agents, running on Solana, allowing developers to configure independent wallets and spending functions for agents without surrendering private keys. From payment protocols to custody solutions to dedicated stablecoins, the industry is advancing on multiple fronts, and this openness is accelerating the formation of the AI agent commercial ecosystem.
What Deeper Impacts Will the Era of AI Agent Commerce Bring
The rise of the AI agent economy is driving a fundamental shift from "human-designed payments" to "machine-designed economic infrastructure." This transition raises several critical questions for the industry. First, responsibility: If an AI agent makes a poor purchasing decision, who is liable? Zabaneh admitted, "This is absolutely a question our entire industry must seriously consider." Second, regulatory adaptation: Traditional payment regulatory frameworks (like PSD2’s strong customer authentication requirements) weren’t designed with AI agents as payment initiators in mind. Whether an AI agent’s strong cryptographic signature can be legally recognized as "valid authorization" remains unclear. Third, trust mechanisms: In open networks, AI agents rely on aggregated identity verification and reputation signals to safely conduct economic activity, but the current lack of identity systems and composite vulnerabilities pose major risks. Finally, as Widmann asked, "How do we integrate agents into existing capital markets and infrastructure?" While crypto payment rails provide the starting point for AI agent commerce, expanding into broader asset management and capital allocation will require integrating agents into the wider financial market system. From payments to assets, from trust to regulation, the era of AI agent commerce is just beginning.
Summary
The remarks by PayPal and Google Cloud executives at Consensus are not isolated statements—they reflect a keen recognition of structural changes in AI agent commerce. The inherent exclusion of AI agents from traditional bank accounts has forced a new logic: Crypto payment rails, with their machine-readable interfaces, real-time settlement, programmability, and global circulation, are deeply aligned with AI-native commerce. From Widmann’s multi-party custody security solution to Zabaneh’s focus on merchant readiness, from the construction of the AP2 open protocol ecosystem to the coordinated efforts of AWS, Exodus, and other industry players, the foundational payment infrastructure for the AI agent economy is developing on multiple levels. With 98.6% of AI agent transactions already choosing stablecoins for settlement, this self-reinforcing trend is accelerating, signaling that crypto payment rails will play a pivotal role in the future of the AI economy.
Frequently Asked Questions
Q: Why can’t AI agents use traditional credit cards or bank transfers for payments?
AI agents lack legal or natural person status, so they cannot meet banks’ KYC requirements or independently apply for multi-currency bank accounts in cross-border scenarios. Traditional payment interfaces are designed for human interaction, lacking standardized programmatic interfaces for autonomous agents. Crypto wallets are managed via private keys, require no account approval, and are inherently suited for machine autonomy.
Q: What is Google’s AP2 protocol?
AP2 (Agentic Payments Protocol) is an open payment protocol for AI agents launched by Google and donated to the FIDO Foundation. Over 120 partners, including PayPal, are participating. The protocol aims to provide a unified, machine-readable payment standard for AI agents to seamlessly transact across platforms.
Q: How is the security of AI agent payments ensured?
Multi-party custody is the industry consensus for security. AI agents do not hold the full private key, only a shard; transactions require signatures from multiple authorized parties to execute. Google has extended its cloud key management system to crypto custody, deploying this architecture accordingly.
Q: What is the projected scale of AI agent-driven commercial activity?
Industry estimates suggest that by 2030, agent-driven commerce led by AI could reach $3–5 trillion. Data already shows that in just the past nine months, AI agents have autonomously completed 140 million crypto payments totaling $43 million.
Q: What are the regulatory and compliance challenges for AI agent payments?
Currently, there is no dedicated regulatory framework for AI agent payments. In the EU, AI agent payments must comply with PSD2’s strong customer authentication requirements, meaning whether cryptographic signatures from AI agents are legally recognized as valid authorization remains to be clarified. Additionally, if an agent makes a poor purchasing decision, liability is not yet addressed in legal frameworks and requires industry-wide exploration.
Q: How should merchants prepare for the AI agent commerce era?
According to PayPal’s survey, only about 20% of merchants currently have machine-readable product catalogs. Merchants need to adjust product information formats to agent-readable data structures (such as structured API interfaces and standardized product catalogs), enabling AI agents to autonomously discover, evaluate, and purchase products. Merchants should also consider automating reconciliation, payment verification, and backend integration in transaction processes to support high-frequency agent transactions.
