The Scale of AI Payments
Alipay, the digital payments arm of Ant Group, disclosed on its official channels that it has processed 300 million AI-powered transactions (referred to as 'AI付'). This milestone positions the platform as the first globally to deploy a large-scale, commercially operational AI-native payment infrastructure. Additionally, Alipay stated that its system now supports 95% of general-purpose AI agents currently available in the market. The company simultaneously launched two new products: an AI wallet designed for agent-managed assets and a Token Pay service that enables autonomous token-based settlements.
These figures come from a public filing by Alipay on the AIbase.cn directory, which aggregates AI industry news and product updates. While the exact time frame of the 300 million transactions was not specified, the sheer volume indicates a rapid adoption curve for AI-driven payment flows, likely spanning both consumer-facing agent interactions and enterprise-level autonomous transactions.
What is Token Pay and the AI Wallet?
Token Pay is a new settlement layer that allows AI agents to initiate and complete payments using cryptographic tokens without direct human intervention. According to Alipay's announcement, Token Pay is built on existing digital payment rails but abstracts away the need for manual authorization in predefined agent workflows. This is especially relevant for scenarios where an AI agent must purchase cloud compute, subscribe to APIs, or compensate other agents for services rendered. The AI wallet, meanwhile, acts as a custodial account managed by an agent's identity rather than a human user. Alipay claims the wallet can hold fiat-pegged stablecoins, loyalty points, and tokenized assets, and it supports programmable spending limits that the agent itself can optimize based on task requirements.
The integration with 95% of general-purpose AI agents suggests that Alipay has developed a universal agent authentication and payment protocol. This likely leverages API-based agent identification, where each agent receives a unique cryptographic identity that is verified during transaction authorization. For comparison, most existing payment systems still require human-in-the-loop confirmation for every transaction, which would bottleneck high-frequency agent interactions.
Why This Matters: Infrastructure for the Agent Economy
The announcement signals a fundamental shift in how digital payments are conceptualized. As AI agents become autonomous - booking travel, managing supply chains, or executing trades - they need financial agency separate from their human operators. Alipay's move directly addresses this friction. The 300 million transaction count indicates that such agents are already operating at scale in China's digital ecosystem. 'AI付' transactions likely include everything from an agent paying for cloud GPU usage to micro-payments between AI bots in a decentralized workflow.
Support for 95% of general-purpose agents means Alipay has likely partnered with major agent platforms like Dify, Coze, or homegrown Chinese equivalents to embed its payment SDK. This creates a network effect: the more agents that use Alipay's infrastructure, the more attractive it becomes for new agent platforms to integrate it. Ant Group is betting that the next wave of digital commerce will be agent-to-agent, with humans only stepping in for high-value exceptions.
Technical and Competitive Implications
From a technical standpoint, Alipay's AI-native payment infrastructure likely relies on agent identity verification via token-based authentication, smart contract on settlement, and real-time risk scoring adapted to agent behavior patterns. The company has long been a leader in risk management for human payments; extending that to non-human actors requires new models for detecting anomalous agent behavior - such as a sudden spike in transaction frequency that could indicate a hacked agent.
Competitively, this positions Alipay ahead of other payment giants. WeChat Pay has yet to announce similar AI-native features. Globally, Stripe and PayPal have begun exploring agent payments, but neither has disclosed transaction volumes in the hundreds of millions. Alipay's head start in China's high-density digital economy gives it a unique testbed. Moreover, by launching an AI wallet, Alipay is effectively creating a new asset class - agent-held balances - that could eventually be used for inter-agent lending, staking, or automated treasury management.
Looking Ahead: Challenges and Opportunities
Despite the impressive numbers, several questions remain. First, who bears liability when an AI agent makes a fraudulent or erroneous payment? Alipay's existing dispute resolution system is designed for humans; adapting it to agents will require new legal frameworks. Second, the 300 million figure may include many low-value, repetitive micro-transactions - such as agent polling fees - which could inflate the count. Nevertheless, the operational experience of handling 300 million agent transactions provides invaluable data for refining reliability and cost.
Looking forward, we expect Ant Group to open the AI wallet and Token Pay to third-party developers via APIs, similar to how Alipay's payment APIs are accessible to merchants. This could spawn a new ecosystem of 'agent finance' (AgentFi) tools. For developers in the AI tools space, this development means that building monetizable agent applications is now infrastructure-level possible in China. The rest of the world will likely watch closely: if agent-to-agent payments become mainstream, every major payments company will need to follow suit.
댓글