ME News

wrote a column ·

No fierce battle in AI payments—can even the tech giants not spark momentum?

Author: Chen Dengxin

Source: Zinke Times

The AI revolution is entering a pivotal leap.

In June 2026, major internet companies such as JD.com, Tencent, and Alibaba all began exploring AI-powered payments, aiming to bridge the 'last mile' of AI commercialization.

After all, the age of intelligent agents demands a closed-loop transaction system.

Yet, nearly a month into this AI payment race, the scene remains surprisingly calm—far from the fierce battle many anticipated. Even as tech giants spar from afar, they have stirred little more than ripples.

Why is AI payment generating more thunder than rain?

The ceiling for traditional payments is now clearly visible.

According to iResearch, China’s personal mobile payment transactions reached 139.078 billion in 2025, up 1.1% year-over-year, while the personal mobile payment market size totaled RMB 205.2 trillion, down 3.7% year-over-year.

Source: iResearch

This means thatThe traditional payments market is nearing saturation and urgently needs new growth drivers.。

Lin Jian, General Manager of Xingyi Pay, stated: 'With the advancement of AI technology, stakeholders connected through payments stand at a historic inflection point, poised for both user experience optimization and business model transformation.'

Coincidentally, AI is also seeking the right partner for real-world implementation.

In previous years, the industry equated parameter scaling in large models with emergent intelligence, fueling an obsession with competing over model size—causing parameters to balloon from tens of billions to hundreds of billions, trillions, and even tens of trillions.

The question is,Real-world implementation is the core logic of large models; for AI to evolve from usable to truly useful, it cannot do without intelligent agents.。

Jensen Huang, founder and CEO of NVIDIA, stated on stage at GTC Taipei 2026: 'Useful AI has arrived—the era of agent-based AI is fully unfolding.'

Against this backdrop, intelligent agents have become the super gateway to AI traffic. Major internet players are ramping up investments to seize greater influence in the industry and secure a higher position in the AI ecosystem.

As shown above, the integration of AI and payments is a natural fit.

More importantly, in 2026 AI entered real-world task scenarios, with agents like Doubao, Yuanbao, and Qwen now capable of executing a range of daily tasks on behalf of users—bringing 'one-sentence shopping' to center stage.

AI enters real-world task scenarios

Take Qwen as an example: within six months of integrating payment capabilities into its AI conversations, it processed 300 million agent-initiated payments and supports 95% of general-purpose agent frameworks.

WeChat Pay officially stated: 'Equipping intelligent agents with full payment service capabilities is a critical step in expanding AI agents from efficiency tools to consumer-facing applications.'

As a result, user behavior has undergone subtle shifts.

As users gradually become accustomed to fulfilling their daily needs through conversations with AI, the payment step becomes the 'last mile.' Driven by the convergence of user demand and technological advancement, AI-powered payments are no longer optional—they are essential.

In short,AI-powered payments complete the commercial loop of 'shopping with a single sentence,' achieving end-to-end integration of the payment chain.。

"Singularity Research Lab" stated: "AI payments are not a top-down invention from big tech firms, nor a meaningless battle for entry points, and certainly not a 'pseudo-demand' that users don’t need. Instead, it is an infrastructure upgrade driven jointly by technology and evolving user habits—a natural evolution for the payments industry from 'transaction processing' to 'intelligent services.'"

Against this backdrop, internet giants are all eyeing AI-powered payments, though their approaches differ.

For example, Alibaba launched an AI-powered version of Alipay, consolidating previously fragmented services into a single conversational interface and using payments as a fulcrum to build a new AI-driven commercial ecosystem.

In short,The underlying logic of payments has evolved from 'people seeking services' to 'services finding people.'。

Payments have entered the era of conversation.

As shown above, Alibaba’s AI-powered Alipay was not developed overnight.

According to LatePost: 'In terms of specific integration methods, Alipay has adopted a “dual-track” approach: on one hand, it encourages willing merchants to proactively integrate by building their services into AI-callable MCPs or Skills; on the other hand, with user authorization, AI performs “screen-reading” operations to interact directly with mini-programs that have not yet been upgraded.'

Similarly, Tencent has launched an AI-dedicated card embedded within WeChat Wallet—a payment feature specifically designed for AI agents. It is already integrated with the WorkBuddy agent and covers high-frequency payment scenarios such as online shopping, food delivery, and afternoon tea.

It follows a dedicated-use-only approach.

Tang Daosheng, Senior Executive Vice President of Tencent Group, stated: 'User habits have continuously evolved over the past 30 years, so our product teams are fully embracing these changes. For example, Tencent Docs’ recent iterations over the past few months have essentially transformed its accumulated document-processing capabilities into Skills—APIs callable by WorkBuddy—enabling office users to access years of refined document-handling functionality, now operating within the new agent-based interface.'

Notably, the AI-dedicated card differs in strategy from Alipay’s AI-powered version.

The former functions as a virtual account, isolated from the primary account and subject to spending limits, placing greater emphasis on security; the latter has fundamentally restructured its service distribution logic, prioritizing its role as a super entry point.

It is evident that the AI-dedicated card takes a more conservative stance, elevating security to the top priority in AI payments, whereas Alipay’s AI version ventures into deeper waters, treating AI payments as a pivotal enabler for intelligent agents and adopting an open posture toward Agent-native ecosystems.

AIX Finance noted: 'When Taobao launched in 2003, nobody dared to send money directly to strangers—escrow transactions served as the intermediary. Today, the AI Pay protocol is doing the same thing: using rules and standards to secure the foundational infrastructure for the next generation of payment entry points. A step too late might mean losing even the chance to join the table.'

Nevertheless, AI payments cannot be achieved overnight, for three main reasons.

First, the issue of trust remains unresolved.。

In retrospect, payments are not merely a technical issue but one of trust—especially against the backdrop of AI’s tendency toward hallucinations and misunderstandings of common sense, making complete reliance on it unwise.

Therefore, even if AI successfully enables high-frequency, low-value commercial scenarios, final payment confirmation still requires human intervention. This significantly diminishes the practical value of AI-powered payments, leading some users to view it as redundant—preferring instead to shop directly on Taobao or pay via WeChat Pay.

Put simply, the inability to complete a transaction seamlessly in one go is AI payments’ biggest weakness.

Secondly, cross-platform integration faces limitations.。

For various reasons, payment interfaces developed across different platforms long remained incompatible, creating a payment fragmentation problem. Although this issue has largely been resolved, it has reemerged in the context of AI payments—given the substantial commercial interests involved, resolving it will take time.

This means that, in the short term, AI payment systems will continue to face cross-platform restrictions.

Thirdly, liability allocation remains ambiguous.。

Although mainstream AI payment systems have implemented various self-imposed safeguards to alleviate user concerns, ambiguities in liability allocation persist and require further refinement.

For example, if an item of clothing is purchased at a price deemed too high and the AI payment system has already automatically deducted the amount, how should the issue be resolved? Since perceptions of what constitutes 'too expensive' vary among individuals, how can fairness be balanced?

Another example: How can AI payments distinguish whether a 'refund-only' request is made in good faith or maliciously? And when faced with fraudulent refund-only claims, how can merchants’ rights be protected?

In short, payment methods are continuously evolving—from QR code payments to facial recognition payments, then to tap-to-pay, and now arriving at AI-powered payments.Although AI-powered payments are still in their infancy and face multifaceted challenges related to trust, regulations, and legal frameworks, making it impossible to satisfy all users,this race waits for no one. After all, entering the game only after the infrastructure is fully built would be too late—and far more costly. No one wants to become that cost.

Thus, even if AI-powered payments haven’t been fully proven yet, the battle over payment entry points must still be fought.

This article originally appeared on the WeChat official account “Xinke Du” (ID: znkedu), authored by Chen Dengxin, and is republished here with permission from 36Kr.

58K Views