The Age of AI Payment Has Arrived: How “Speak-to-Pay” Is Reshaping China’s Payment Ecosystem

1. The Milestone Month: June 2026

If there was ever a moment that marked the official arrival of AI payment in China, it was June 2026. Within a single week, the country’s leading tech and payment giants unleashed a cascade of announcements that fundamentally redefined how payments would work in the AI era.

What made this week extraordinary wasn’t just the number of announcements — it was the convergence. After years of incremental AI integration in payment systems, every major player suddenly moved in unison, signaling that the industry had reached an inflection point.

2. How AI Payment Works: From Human-Initiated to Agent-Driven

To understand the significance of these developments, it helps to first understand the fundamental shift in how payments are initiated and executed.

This shift is profound. In the traditional model, the user is the operator — every action requires manual input. In the AI payment model, the user becomes the director — setting the intent and boundaries, while the agent handles execution[reference:19].

JD.com’s A2P2 protocol provided the most systematic framework for understanding this evolution. The six-level classification — from L0 (full human confirmation) to L5 (full agent autonomy) — maps out a clear progression[reference:20]. The protocol focused particularly on levels L3 and L4: L3 allows agents to initiate payment requests within a single task, with the system deciding whether to approve based on user-defined boundaries; L4 grants agents more authorization to complete payments autonomously as long as the amount, scenario, and user credentials fall within preset parameters[reference:21].

The core innovation of A2P2 is the “Mandate” mechanism — converting a user’s natural language instruction (e.g., “buy a bouquet of flowers for under 200 yuan for my friend”) into a machine-verifiable task credential[reference:22]. The system then checks the amount, category, and payee in real time before approving or rejecting the payment[reference:23].

Security, of course, is paramount. The A2P2 protocol introduced the ARI (Agent Runtime Identity) mechanism, which binds three elements in real time at the moment of payment: the actual user, the agent’s identity, and the agent’s runtime environment[reference:24]. If an agent is compromised and attempts an unauthorized transfer, ARI detects the runtime anomaly and blocks the request immediately[reference:25].

3. Major Players’ AI Payment Strategies Compared

While all major players are moving toward AI payment, their approaches differ significantly based on their existing ecosystems and strategic priorities.

Alipay: The All-In AI Super App

Alipay’s approach is the most ambitious. Rather than adding an AI feature, it rebuilt the entire app around AI. The traditional nine-grid interface has been replaced by a minimalist dialog box as the primary entry point[reference:42]. Users can access tens of thousands of services — from bill payments and transportation to retail and financial queries — simply by speaking or typing a command[reference:43].

Behind the scenes, Alipay has built what it calls a “full-stack AI-native payment infrastructure,” covering authorization, payment, settlement, security, and trust[reference:44][reference:45]. As of May 2026, the platform had processed over 300 million AI agent payment transactions and was compatible with 95% of mainstream general-purpose large model agent frameworks[reference:46][reference:47].

In July 2026, Alipay took the next step by launching its AI Open Platform, allowing merchants, institutions, service providers, smart terminals, and large model platforms to access its AI capabilities[reference:48][reference:49]. Early adopters include KFC, Mixue Bingcheng, Luckin Coffee, Amap, Didi, and other major brands across dining, travel, and lifestyle services[reference:50]. As one observer put it: “May solved whether AI can pay. June solved where users enter. July solved where merchants enter and how they get called.”[reference:51]

WeChat Pay: Security-First, Lightweight Integration

WeChat Pay’s AI Exclusive Card takes a fundamentally different approach. Rather than rebuilding the app, it creates a separate, isolated virtual fund pool within the existing WeChat wallet[reference:52][reference:53]. The card is completely segregated from the user’s main account — agents can only spend from the Exclusive Card balance, never from the main wallet[reference:54][reference:55].

Security is built on three pillars: account isolation (the card is physically separated from the main account), balance autonomy (users decide how much to transfer in and can adjust at any time), and per-transaction confirmation (every order requires the user’s phone-side approval before execution)[reference:56].

The card was first integrated with Tencent’s WorkBuddy desktop productivity agent. Users can summon the “Meituan Life Assistant” within WorkBuddy, ask for nearby restaurant deals, review recommendations, and then have the agent initiate payment — with the user still confirming on their phone[reference:57]. This approach prioritizes security and control over seamlessness, making it ideal for users who are cautious about giving agents spending power[reference:58].

ChinaUMS: Bridging the Physical World

ChinaUMS brings a unique perspective as a payment service provider deeply embedded in the physical merchant ecosystem. Its AI payment product — launched on the same day as Alipay’s A-Bao — focuses on real-world transaction scenarios where voice interaction can dramatically simplify the user experience[reference:59][reference:60].

The product has been piloted in three high-frequency scenarios: smart campus food reservations, AI food ordering, and cloud payments[reference:61]. For merchants, the value proposition is clear: reduced labor costs and improved operational efficiency, all while maintaining strict identity authentication, end-to-end data encryption, and transaction record-keeping[reference:62].

ChinaUMS has also upgraded its smart payment terminals with AI capabilities, including voice wake-up, voice operation, smart collection, and online Q&A[reference:63]. In May 2026, it launched “Payment Skills” — a tool that reduced the development and integration cycle for payment access from 5-7 working days to just minutes[reference:64].

China Mobile (中移金科): Telecom-Integrated AI Payment

China Mobile’s subsidiary 中移金科 launched its “Mobile Wallet AI Pay” in April 2026, predating the June rush[reference:65][reference:66]. The product is built around three core principles: intent-driven, human-AI collaboration, and trustworthiness[reference:67].

What sets it apart is its deep integration with China Mobile’s telecom ecosystem. The AI Pay product supports voice-command triggered top-ups, bill payments, and other services, reducing page jumps and repetitive input[reference:68]. It also features intelligent asset coordination — automatically integrating wallet balances, bank cards, points, and coupons to recommend the optimal payment mix[reference:69]. On the security front, it employs self-developed AIGC anti-fraud and “Yutong” anti-fraud large models, with millisecond-level risk control and support for iris and voiceprint biometric payments[reference:70]. The platform is already connected to over 350 banks and covers more than 50 million offline merchants[reference:71][reference:72].

4. The Business Significance: From Traffic to Intelligence

The shift to AI payment is far more than a user experience upgrade. It represents a fundamental transformation of the payment business model.

Payment Entry Points Are Changing

For the past decade, the payment industry’s growth engine was traffic — acquiring more users, more merchants, more transactions. The primary battleground was the “scan” — QR codes, barcodes, NFC taps. Revenue was concentrated in transaction channel fees, a model under increasing margin pressure[reference:73].

AI payment changes the game. The entry point shifts from “scan” to “speak” — from the camera to the microphone. The competitive advantage moves from who has the most merchants to who has the most capable AI agent ecosystem.

As one industry observer noted: “The race is no longer for the payment entry point, but for the first entry point between users and services in the AI era.”[reference:74] Alipay’s A-Bao and WeChat’s AI capabilities are both positioning for this — not just to process payments, but to become the primary interface through which users access all services[reference:75].

New Revenue Streams Emerge

The AI payment ecosystem is opening up entirely new revenue channels. Analysts point to several emerging opportunities[reference:76]:

• Merchant digital services: As AI agents become the primary interface for service discovery and purchase, merchants will pay for visibility and integration within AI ecosystems.
• AI industry settlement: Large model providers and AI service platforms need payment infrastructure that supports token-based purchasing and automated settlement[reference:77].
• Smart terminal hardware: AI-compatible payment terminals, voice-interactive devices, and lightweight security chips are seeing growing demand[reference:78].
• Third-party risk services: Smaller payment institutions and regional merchants lack the resources to develop their own financial AI models, creating demand for third-party fraud detection and behavioral prediction services[reference:79].

Consumer Readiness Is High

The technology is one thing — but are consumers actually ready for AI agents to spend their money? The data suggests yes.

Another survey found that 84.56% of Chinese consumers have already tried AI-assisted shopping features, and over 85% are looking forward to one-stop AI shopping experiences[reference:82]. Nearly 40% are “very eager” for one-click agent shopping that simplifies the entire process[reference:83].

These numbers suggest that Chinese consumers are not just ready for AI payment — many are already actively using it. The infrastructure is being built to meet demand that already exists.

5. What This Means for International Businesses

For international enterprises, law firms, financial institutions, and investors engaging with China, the rise of AI payment is not a distant trend — it’s a present reality that directly affects how business gets done.

• Your Chinese partners are adopting AI payment. Whether they’re using Alipay’s A-Bao, WeChat Pay’s AI Exclusive Card, or other platforms, the companies you work with are likely already integrating AI payment into their operations. Understanding how these systems work — and their security implications — is essential for risk assessment.
• Due diligence must evolve. AI payment introduces new vectors for both efficiency and risk. When vetting a Chinese partner, consider asking: What AI payment systems do they use? How are agent-initiated transactions authorized and tracked? What security protocols are in place? These questions may not be standard yet — but they soon will be.
• New compliance considerations emerge. The regulatory framework for AI payment is still developing. The Financial Law draft, the Classification Rating Measures, and the Cybersecurity Management Measures all have implications for AI-enabled payment systems. Professional Enterprise Credit Reports can help you stay ahead of compliance requirements by providing comprehensive, up-to-date information on your partners’ regulatory standing.
• Cross-border payment intelligence is coming. The same AI capabilities that enable domestic “speak-to-pay” will eventually extend to cross-border transactions. This will reduce friction for international commerce — but also require new approaches to currency conversion, compliance, and fraud prevention.

The AI payment revolution is not coming — it’s already here. June 2026 was the moment the industry crossed the threshold. The companies that understand this shift — and prepare for its implications — will be the ones that thrive in the next era of Chinese commerce.