Peng Jiayan/Text

On April 10, Hong Kong issued its first batch of stablecoin licenses, but JD, which had already entered Hong Kong's stablecoin sandbox and was fully prepared except for the license, did not receive one.

The industry had long speculated about JD's exclusion, but when the decision was finalized, looking at another project recently launched by JD—ClawTip—might offer deeper insight into the strategic value of this license for JD.

On March 31, 2026, JD Technology introduced ClawTip, an A2A (Agent-to-Agent) micropayment infrastructure for the AI Agent ecosystem, built on the x402 protocol.

In the evolution of technological paradigms, the rise and fall of commercial giants often hinge on their control over underlying infrastructure. As artificial intelligence advances from "content generation" to "executing commercial tasks," the agent economy is giving rise to a new generation of financial payment networks.

Stablecoins + x402 are considered one of the most suitable combinations for agent payments. If JD had a stablecoin license, it could form a "money issuance + payment channel" strategic framework with ClawTip.

After missing out on the mobile payment dividends of the past fifteen years and adopting a defensive stance in the large model race in recent years, ClawTip represents JD's attempt to enter the agent economy, but the absence of a stablecoin license leaves it at a disadvantage.

In this fiercely competitive agent economy, can ClawTip become JD's vanguard for a comeback?

01

The "Historical Debt" of the Mobile Payment Era

To understand JD's aggressive foray into AI payments, one must revisit its strategic passivity during the mobile payment era. Over the past decade, the market landscape for third-party payments in China has solidified.

As of 2025, Alipay and WeChat Pay collectively dominate over 90% of the market share. This means that during the heyday of mobile payments, JD never managed to gain traction, with its payment business serving primarily as a supporting tool for its e-commerce ecosystem rather than as an independent market force.

The far-reaching chain reaction (ripple effects) of this structural lag have left JD in a follower position in consumer (C-end) payment infrastructure competition, failing to establish a foundational financial moat.

In 2025, JD Group's annual revenue reached RMB 1,309.1 billion, up 13% year-on-year, with annual active users surpassing 700 million. Its core e-commerce business remains robust.

However, with payment entry points firmly controlled by competitors, JD's imagination in financial services has always been limited. Not to mention the widening gap with Alipay and WeChat Pay in fundamental capabilities like user behavior insights and credit risk control, which rely on payment data.

02

Failure to Secure a Hong Kong Stablecoin License

If mobile payments represented JD's first missed wave, stablecoins should have been its second opportunity for a breakthrough.

With the maturation of Web3 technology, stablecoins are widely seen as the "numéraire" for next-generation cross-border trade and the agent economy.

In 2024, JD's wholly-owned subsidiary, JD Coinlink Technology (Hong Kong) Limited, was selected for the Hong Kong Monetary Authority's first batch of "stablecoin issuer sandbox" participants.

Liu Qiangdong, Chairman of JD Group, explicitly expressed strategic ambitions at an internal sharing session: "JD plans to apply for stablecoin licenses in all major currency countries globally... with the goal of reducing global cross-border payment costs by 90% and increasing efficiency to under 10 seconds per transaction."

On April 10, 2026, the first batch of fiat-backed stablecoin issuer licenses in Hong Kong was officially announced, with only two institutions approved—HSBC and Standard Chartered's joint venture, Dingdian Fintech.

Eddie Yue, Chief Executive of the Hong Kong Monetary Authority, stated, "We maintain an open yet cautious attitude toward issuing additional licenses and the timing. Even if more licenses are issued in the future, the total number will remain very limited."

JD entered the sandbox but failed to secure a license in the first round. This means JD's plan to leverage Hong Kong's stablecoin license to drive global cross-border payment ambitions has suffered a setback.

While JD can still apply for licenses in other jurisdictions—Singapore, the UAE, the EU, and elsewhere are advancing relevant frameworks—Hong Kong's strategic value as a bridge between mainland China and international markets makes its stablecoin license irreplaceable for JD's global ambitions.

03

Lagging Behind in the Large Model Race

Large models have ushered in a new era of the internet economy and serve as the cornerstone of the agent economy. On this critical and capital-intensive track for tech companies, JD's position is equally precarious.

Among domestic players, traditional internet platforms like Alibaba, ByteDance, Tencent, and Baidu have internationally competitive models, while a wave of startups has carved out niches in both the model space and capital markets.

Although JD is advancing its large model layout (deployments), its publicly disclosed achievements and applications lag noticeably behind these four competitors.

In 2025, JD released its self-developed AI large model, "Yanxi" (JoyAI), claiming deployment in over 1,000 business scenarios and sustained growth in AI shopping assistant users.

By the end of 2025, JD's cumulative technology investment exceeded RMB 170 billion, with a 66% year-on-year increase in the latest quarter—a substantial commitment.

In March 2026, JD unveiled multiple AI business progressions, including large models, digital humans, and "embodied intelligence" products, such as the Instruct version of its open-source general-purpose foundational model JoyAI-LLM Flash, digital human JoyStreamer, and embodied intelligence JoyInside.

The issue lies in JD's AI strategy, which focuses heavily on "industrial-side" applications and internal efficiency gains rather than exporting influential AI platform capabilities externally.

Yanxi's core use cases revolve around JD's internal e-commerce recommendations, logistics optimization, and customer service automation. While it has accumulated 70,000 merchants using digital humans and nearly 100 collaborating home appliance and furniture brands, its industry influence pales in comparison to Alibaba's Tongyi Qianwen, ByteDance's Doubao, Baidu's Wenxin Yiyan, or Tencent's Hunyuan.

As a company centered on supply chain efficiency, JD's AI strategy is summarized as "the value of AI = model × experience × industrial depth," emphasizing deep integration with physical industries rather than competing on general-purpose model capabilities.

This "pragmatic" approach may be commercially robust , but in the competition for AI-era discourse power, JD lacks a "universal powerhouse" model with top-tier generalization capabilities that can widely reach developers globally.

04

Launch of ClawTip

On March 31, 2026, JD Technology introduced ClawTip, an A2A (Agent-to-Agent) micropayment infrastructure for the AI Agent ecosystem.

It breaks away from traditional payment friction—"human confirmation, password input, facial scan verification"—by granting AI the ability to make "decision-based payments" for the first time.

ClawTip enables AI agents to pay autonomously. Instead of AI "assisting" users in payments and prompting fingerprint confirmation via a popup, it truly empowers AI with independent judgment, decision-making, and transaction completion capabilities.

According to product disclosures and media reports, its core technical features include:

Agent-Exclusive Wallets: Each AI is assigned an independently physically isolated wallet. AI can only operate within preset sandbox limits without accessing users' core bank accounts.

AI Autonomous Decision-Making: Leveraging intent recognition, AI can independently inquire about service prices online (e.g., purchasing computing power or calling third-party APIs), judge the transaction's value, and complete payment.

Trusted Snapshot Evidence: Built-in agent behavior capture ensures every human-independent transaction is traceable and auditable.

ClawTip constructs an A2A interaction protocol and collaboration framework based on the x402 protocol, providing AI agents with a unified "communication language" for cross-platform, cross-system autonomous task collaboration.

This enables AI to automate the entire process from "service discovery → quote acquisition → invocation confirmation → automatic billing and payment → result retrieval," achieving true "pay-as-you-call, settle-as-you-serve."

According to ClawTip's official website, it offers four use cases:

Skill Monetization: Package skills as services for AI to call—automatically collect payments, worry-free and hassle-free.

AI Autonomous Payments: Write reports, query data, purchase services... AI decides, AI pays, intelligent delivery.

Human-Machine Collaboration: Small AI payments, large payments require confirmation—ensuring fund security.

Tipping AI: Encounter useful skills, impressive AI... tip with a single sentence to encourage.

ClawTip announces a "zero-fee model" for developers—completely free payment, collection, and withdrawal—to address monetization challenges faced by independent AI developers (e.g., high barriers to subscription models), thereby rapidly capturing market share in the industry's early stages and establishing a de facto standard for A2A settlements.

05

The Agent Payment Battlefield Surrounded by Giants

Although agent payment is a new track (track), numerous top-tier players have already entered, with more following suit.

In July 2025, Ant Group provided MCP (Model Context Protocol) interfaces, allowing third-party agents to integrate Alipay's payment capabilities and complete the full link (chain) of "agent invocation → service quote → user confirmation → automatic payment."

In February 2026, Alipay disclosed that its "AI Pay" had surpassed 100 million users, with over 120 million cumulative payments in the previous week, becoming the world's first AI-native payment product to exceed both 100 million users and payments.

For WeChat Pay, in July 2025, Tencent Yuanqi officially integrated with WeChat Pay MCP, enabling developers to complete transactions via agent-invoked WeChat Pay. Backed by WeChat's super-ecosystem with 1.3 billion monthly active users, WeChat Pay inherently possesses the broadest user reach in AI payment scenarios.

On April 10, 2026—the same day Hong Kong announced its first batch of stablecoin licenses—21st Century Business Herald reported that WeChat, Alipay, Huawei, and others had launched "payment Skills" to seize payment entry points in the agent ecosystem. This means the competition for AI payment standards has escalated from protocol to application layers.

International innovations are even more diverse.

In fall 2025, Google launched the Agent Payments Protocol (AP2), an open-source agent payment protocol developed in collaboration with over 60 payment and tech giants, including PayPal, Stripe, Ant International, Coinbase, and Mastercard.

AP2 introduces tamper-proof, cryptography-based "authorization letters." It supports traditional credit cards and real-time bank transfers while natively integrating stablecoins and cryptocurrencies, laying the foundation for decentralized machine-to-machine micropayments.

In January 2026, Google unveiled the Universal Commerce Protocol (UCP) at the National Retail Federation (NRF) annual conference, an open standard for AI agent commerce aimed at enabling seamless interoperability among AI platforms, enterprises, and payment providers.

UCP's release indicates Google's ambition extends beyond payments to defining the "commerce protocol layer" of the agent economy.

Payment companies like PayPal and Stripe have partnered with OpenAI to adopt the Agentic Commerce Protocol (ACP), allowing ChatGPT users to complete shopping and payments directly within the chat interface via third-party payment wallets, implemented since 2026.

In March 2026, Stripe and blockchain company Tempo jointly launched the Machine Payments Protocol (MPP), enabling AI agents to autonomously complete payments during task execution, supporting micropayments as low as 0.01 USDC and aligning with agents' high-frequency, small-value usage patterns.

Coinbase, the largest U.S. virtual asset exchange, initiated the x402 protocol in May 2025, collaborating with Cloudflare, Google, Visa, and others to activate the HTTP 402 standard for building an on-chain payment layer for AI agents.

In February 2026, Coinbase launched Agentic Wallets, enabling AI agents to independently hold funds, make payments, and trade tokens, prioritizing AI agent payments as one of the company's top initiatives.

Stablecoin giants Circle and Tether are also building foundational infrastructure for AI payments.

Circle introduced the USDC smart contract layer to support programmable payments; Tether, with a market cap exceeding $200 billion, is exploring applications of stablecoins like USDT in AI agent micropayments.

Mapping these layout (deployments) into a competitive landscape reveals three distinct tiers in the AI payment field: Ant and WeChat dominate domestic C-end AI payments; Google, PayPal, and Stripe compete for agent commerce entry points in the global internet payment ecosystem; Coinbase, Circle, and Tether build AI-native payment infrastructure on the crypto payment track.

From birth, ClawTip faces a breakout challenge.

06

Conclusion

Our economic system is introducing a new class of participants. Agents are transforming from information-processing tools into executing entities capable of invoking services, coordinating resources, and even directly participating in payment flows.

Just as Alipay's escrow payments started (ushered in) the e-commerce era and quick payments started (ushered in) the mobile payment era, AI payments—or A2A payments—are open (ushering in) the agent economy era.

AI payments have become a battleground. JD attempts to gain an early advantage at the AI payment protocol level.

If ClawTip can first establish a sufficiently large network effect within the developer ecosystem, JD may accomplish in the agent economy era what it failed to do during the mobile payment era—dominate payment infrastructure.

JD's three battles in payments—mobile payments, stablecoins, and AI payments—reflect a difficult struggle for an e-commerce platform to transcend its genetic limitations.

Stablecoins + x402 are considered one of the most suitable combinations for agent payments.

The absence of Hong Kong's first batch of stablecoin licenses does not mean JD lacks future opportunities, but time is the greatest cost in the agent era. While JD can still apply for stablecoin licenses elsewhere, holding a Hong Kong license might reduce the difficulty and time costs of applications in other regions.

The road of ClawTip's counterattack is just the beginning, far from the end, and everything is possible.

-End-