140 Billion Agents Enter the Fray: The 'Traffic' Moat Is About to Collapse

At the Alipay AI Ecosystem Conference yesterday, Ant Group CEO Han Xinyi set a different tone from typical 'technology changes everything' opening remarks. He emphasized that while AI brings many changes, it ultimately serves people, and humans retain final decision-making power. However, he noted that in the Agent era, agents themselves begin to enter business chains, fundamentally shifting how transactions and decisions are executed.

Han introduced a key insight that could reshape industry rules: in the Agent era, the logic of traffic will fail, but the logic of trust will rise. This view is now becoming an industry consensus globally, with tech and payment giants like OpenAI, Google, Visa, and Mastercard all moving toward trust mechanisms for agent-driven commerce.

**Key Point One: 'Human Traffic' Gives Way to 'Agent Ecosystems'**

Han argued that size and network effects still determine business moats, but what changes is that the 'traffic gateway' once determined by humans will be replaced by 'agent ecosystems.' The scale of agents, their ability to collaborate, and the network effects among them will become more critical competitive barriers than traditional homepages, search boxes, or recommendation feeds. 'In the mobile internet era, traffic was king, and human-generated access traffic was the moat. But in the Agent era, the agent ecosystem will be a bigger moat,' Han said. He added, 'China has 1.4 billion people, but there may be 140 billion agents.'

In the mobile internet era, a platform's value largely depended on human access traffic. In the Agent era, the entry point will shift from visible pages to the task scheduling network behind agents. Users will no longer complete all steps themselves; instead, they will describe their needs to an agent, which will understand intent, plan tasks, invoke other agents, merchant systems, data services, and payment networks. A new network effect will form around agents—how many services an agent can access, how many systems can recognize it, how many other agents can collaborate securely, and whether it can reliably complete tasks. This is the agent ecosystem Han described.

Global AI and tech giants are already moving in this direction. Google launched the A2A (Agent-to-Agent) protocol to enable agents from different frameworks and platforms to communicate, exchange information, and coordinate actions. OpenAI and Stripe introduced the ACP (Agentic Commerce Protocol), upgrading ChatGPT from a Q&A gateway to a transaction gateway. Alipay itself launched China's first ACT protocol earlier this year, now upgraded to version 2.0 with 20 partners, covering AI-native companies, smart hardware, automotive, payment firms, and retailers, systematically building A2A and A2M (Agent-to-Machine) payment frameworks.

**Key Point Two: Agents Restructure Business Decision-Making**

The second level of change occurs in the decision-making chain. Han stated that agents will restructure decision rights from human-only to joint human-agent decision-making. This leads to three paradigm shifts: First, from 'people finding services' to 'services finding people.' Users only need to express goals, and agents autonomously understand, decompose, match, schedule, and deliver the best service. Trust becomes the core metric for distribution, replacing traffic exposure and click-through rates. Second, from 'product/service transactions' to 'task transactions.' Instead of separately booking flights, hotels, and ride-hailing for a trip, users give an overall goal, and agents deliver a complete solution. Merchants must make their goods and services understandable and embeddable into agent task chains. Third, from 'bilateral markets' to 'multilateral networks.' Besides users and merchants, agents, AI model providers, payment networks, data services, and developers all become nodes in a highly collaborative network, creating new scenarios and efficiencies but also challenges in trust, accountability, data security, and standards.

**Key Point Three: AI Payment Becomes a New Global Infrastructure**

If agents restructure the entire transaction process, payment must also be redefined. Han argued that AI payment will evolve from a payment tool to a new infrastructure supporting the agent-driven commercial ecosystem. Two major transformations are needed: First, the execution entity changes from human to agent. This requires a traceable, auditable chain of responsibility for agent transactions—a precondition for autonomous agent commerce. Second, the value carrier expands from fiat currency to tokens. Since agents operate 24/7, transactions become extremely small, high-frequency, and diverse. Tokens can represent points, rights, coupons, digital assets, enabling efficient circulation and settlement.

Global payment networks are already moving: Visa launched Intelligent Commerce with APIs for AI partners; Mastercard unveiled Agent Pay for secure, scalable payment infrastructure; Stripe is also commercializing transaction capabilities for agentic commerce. These moves align with Han's vision that AI payment must address trust, efficiency, and value granularity in the agent era.

**Alipay's Role: Trust Layer, Connector, Enabler**

Han positioned Alipay's role in the Agent era as a three-layer structure: a trust layer for reliable execution, a connector for collaboration among consumers, agents, and merchants, and an enabler to help both large model firms and small merchants enter the agent economy. This echoes Alipay's historical evolution—from escrow payment in the e-commerce era to mobile payment in the mobile internet era, and now to AI payment in the agent era. The new question Alipay aims to answer is: when transactions are executed by agents, how does the commercial system remain trustworthy, controllable, and settleable?

In summary, Han's talk was not only a strategic statement for Alipay but also a clear summary of the global consensus on the agent economy: competition is shifting from single traffic gateways to collaboration, trust, and executable capability within agent ecosystems. Those who can be discovered, trusted, invoked, and settled by agents will be closer to the next round of commercial distribution. A new business paradigm is emerging.