Summary
Tencent is negotiating to become the largest external shareholder of Agentic AI pioneer Manus, filling the gap left after Chinese regulators blocked Meta's deal. Meanwhile, Meta's Chief Data Officer says autonomous AI agents executing commercial transactions will be core to future business, with stablecoins already the default payment rail inside Meta.
The Convergence of AI Agents and Payments: Two Tech Giants, Different Paths
In the rapidly evolving landscape of AI agents in 2026, two seemingly independent news events reveal the intense competition among tech giants in the Agentic AI space. According to Bloomberg, Tencent is negotiating to become the largest external shareholder of Agentic AI pioneer Manus, a move seen as filling the investment gap left after Chinese regulators blocked Meta's previous attempt to acquire or invest in the company.
Simultaneously, Meta's Chief Data Officer Alex Schultz detailed Meta's vision for Agentic Commerce in a CoinDesk Spotlight interview, explicitly stating that stablecoins have become the default payment rail inside Meta. Together, these developments sketch a clear trend: AI agents will not only transform human-computer interaction but will fundamentally restructure the underlying logic of commercial transactions and payment settlements.
The Battle for Manus: Strategic Positioning by Chinese and American Tech Giants
Manus, as a pioneer in the Agentic AI field, holds strategic value through its core technology enabling AI agents to autonomously execute complex tasks. Meta's previous attempt to gain access to Manus's capabilities through investment or acquisition was blocked by Chinese regulators, highlighting the compliance challenges facing cross-border tech investments in the current geopolitical environment.
Tencent's current negotiations to become Manus's largest external shareholder represent not just strategic compensation for Meta's setback but also reflect the long-term positioning of Chinese tech giants in the AI agent domain. For Tencent, Manus's technological capabilities could synergize with its massive social ecosystem (WeChat, QQ) and payment infrastructure (WeChat Pay), particularly in scenarios where AI agents need deep integration with payment systems.
Notably, this deal remains in negotiation stages, and whether it will ultimately close depends on regulatory attitudes and negotiations on key terms including valuation and control. Regardless of the outcome, Manus has become a critical target in the global tech giants' race for Agentic AI dominance.
Meta's Agentic Commerce Vision: From Social to Autonomous Transactions
The Agentic Commerce vision articulated by Meta's Chief Data Officer Alex Schultz essentially elevates AI agents from assistive tools to autonomous transaction entities. He revealed that Meta currently has over one million businesses using Meta AI agents on a weekly active basis, a number that was essentially zero at the start of the year, demonstrating rapid enterprise adoption of AI agent capabilities.
Schultz used a deliberately mundane example to illustrate Agentic Commerce's potential: coordinating a child's birthday party. AI agents booking times, checking calendars, finding venues, communicating with other parents' agents, all on WhatsApp. The key to this mundane example is its scalability—if agents can handle low-stakes logistics, they can handle supply chain negotiations, financial settlements, and cross-border commerce.
"You write that example large," Schultz said, "and then if you're us, you hope that you do it over WhatsApp." This statement reveals Meta's strategic intent: transforming its massive social platforms into infrastructure for AI agents to execute commercial transactions.
Stablecoins: The Default Rail for AI Agent Payments
Schultz explicitly stated that in Meta's vision, physical wallets will become obsolete, with digital payments representing the entire future. More importantly, stablecoins have already become the default payment rail inside Meta. He pointed out that Meta completely believes in a future with no physical wallets, with digital payments becoming everything, specifically referencing WeChat's red envelope model.
This statement carries significant implications. For AI agents, traditional payment authorization processes (such as entering passwords or biometric verification) create obvious friction. If agents require human authorization for every transaction, the value of "autonomous" transactions diminishes substantially. Stablecoins, as native digital assets, can enable smoother automated payments while maintaining security through mechanisms like smart contracts, multi-signature schemes, and spending limits.
However, Schultz also acknowledged that while stablecoins are already a given inside Meta, the real challenge is "getting the rest of the world there." This involves a complex array of issues including regulatory compliance, user education, and cross-border settlement.
The Core Challenge of AI Agent Payments: Authorization and Compliance
In the Agentic Commerce vision, the greatest technical challenge is not AI capability itself but how to enable AI agents to safely and compliantly obtain payment authorization and execute settlements. This involves several critical questions:
First is the authorization mechanism. Users need to establish clear authorization boundaries: which types of transactions can be executed automatically? What are the per-transaction amount limits? Which transactions require human confirmation? This requires balancing user experience convenience with security.
Second is identity verification. When AI agents execute transactions on behalf of users, how do we ensure the agent itself has not been hijacked or tampered with? This may require combining technologies like zero-knowledge proofs and trusted execution environments.
Third is compliance. Different jurisdictions have varying regulatory requirements for automated payments, AI decision-making, and cross-border fund flows. AI agents need to understand and comply with these rules, placing higher demands on their "intelligence."
For institutional applications, these challenges are even more pronounced. When enterprises use AI agents to execute procurement, settlement, and other business operations, they require stricter permission management, audit trails, and risk control mechanisms. This is why institutional-grade wallets and custody solutions will play a critical role in the Agentic Economy—they need to provide infrastructure that supports both automation and compliance requirements.
From Social Red Envelopes to Global Settlement: The Evolution of Payment Infrastructure
Schultz's reference to WeChat's red envelope model was no accident. WeChat red envelopes succeeded by socializing and gamifying payment behavior, lowering users' psychological barriers. More importantly, it proved that within a closed ecosystem, digital payments can completely replace cash and become the default choice.
Agentic Commerce aims to expand this model globally and extend it from C2C micro-payments to B2B large-scale settlements. This requires solving several key issues:
First is cross-border interoperability. Different countries and regions have varying payment systems, stablecoin standards, and regulatory frameworks, and AI agents need to seamlessly switch between these heterogeneous systems.
Second is liquidity management. When AI agents need to execute transactions across multiple currencies and multiple chains, how do we ensure optimal exchange rates and minimal slippage? This may require integrating DeFi protocols, market maker networks, and other liquidity sources.
Third is dispute resolution. When transactions executed by AI agents encounter problems (such as undelivered goods or services not meeting expectations), how do we arbitrate and process refunds? This requires establishing new dispute resolution mechanisms, potentially involving innovations like smart contract escrow and decentralized arbitration.
The Future of the Agentic Economy: From Assistance to Autonomy
Schultz quoted science fiction author William Gibson's famous line to describe the agentic economy: "The future is already here, it's just not evenly distributed." This formulation accurately captures the current development stage of Agentic AI—the technology is already viable, but widespread application still faces numerous obstacles.
Meta's current strategy is to start with low-risk, high-frequency scenarios and gradually build user trust in AI agents. From coordinating birthday parties to booking restaurants to handling customer service, these scenarios share the common characteristics of relatively simple decision-making and larger error tolerance. But as AI capabilities improve and user trust builds, agents will gradually enter more complex, higher-value domains.
For payment and financial service providers, this trend means rethinking product design. Traditional payment products assume users are active decision-makers, while products for the AI agent era need to support collaboration between "passive" users and "active" agents. This is not just a technical challenge but a challenge for business models and risk management.
The Institutional Perspective: Infrastructure for Autonomous Commerce
While much of the current discussion around AI agents focuses on consumer applications, the institutional implications are equally profound. For businesses managing treasury operations, procurement, and cross-border settlements, AI agents represent both opportunity and risk.
Institutional adoption of AI agents for financial operations requires infrastructure that can handle complex authorization hierarchies, real-time compliance checks, and comprehensive audit trails. This is where institutional-grade custody and wallet solutions become critical. These systems need to support programmable authorization rules that allow AI agents to operate within defined parameters while maintaining human oversight for exceptional cases.
The challenge is particularly acute in regulated industries where fiduciary duties, anti-money laundering requirements, and sanctions compliance cannot be delegated to autonomous systems without appropriate safeguards. Solutions will likely involve hybrid approaches where AI agents handle routine operations while flagging edge cases for human review, with all actions recorded on immutable ledgers for regulatory reporting.
Regulatory Considerations: Navigating the Compliance Landscape
The regulatory dimension of Agentic Commerce cannot be overlooked. As AI agents begin executing financial transactions autonomously, regulators worldwide are grappling with fundamental questions about liability, consumer protection, and systemic risk.
Who is responsible when an AI agent makes an unauthorized transaction? How do existing know-your-customer (KYC) and anti-money laundering (AML) frameworks apply when the transacting entity is an AI rather than a human? What disclosure requirements should apply when AI agents are making purchasing decisions on behalf of consumers?
These questions have no easy answers, and different jurisdictions are likely to take different approaches. The blocking of Meta's Manus investment by Chinese regulators illustrates how geopolitical considerations intersect with technology governance. Companies operating in this space will need to design systems that can adapt to varying regulatory requirements across markets.
The stablecoin dimension adds another layer of complexity. While Meta has embraced stablecoins as its default payment rail internally, the regulatory status of stablecoins varies widely across jurisdictions. Some countries have established clear frameworks, while others maintain ambiguous or restrictive positions. For Agentic Commerce to scale globally, greater regulatory clarity and harmonization will be essential.
The Competitive Landscape: Ecosystems vs. Interoperability
The competition between Tencent and Meta over Manus highlights a fundamental strategic question: will the Agentic Economy be dominated by closed ecosystems or open, interoperable protocols?
Meta's vision of conducting Agentic Commerce "over WhatsApp" suggests a platform-centric approach where Meta controls the infrastructure layer. Tencent's WeChat ecosystem demonstrates the power of this model in practice—within China, WeChat has become the de facto operating system for digital life, with payments, commerce, and social interaction all occurring within a single app.
However, the global market may not support multiple closed ecosystems. If AI agents are to coordinate across platforms—a user's Meta agent communicating with a merchant's Shopify agent, for example—some degree of interoperability will be necessary. This could take the form of open standards, cross-platform protocols, or blockchain-based infrastructure that enables agent-to-agent transactions without platform intermediation.
The outcome will likely be a hybrid model, with dominant platforms maintaining significant control over their ecosystems while supporting limited interoperability for specific use cases. The companies that can navigate this balance—maintaining competitive advantages while enabling enough openness to drive network effects—will be best positioned for success.
Looking Ahead: The Path to Mainstream Adoption
For all the promise of Agentic Commerce, widespread adoption will require overcoming significant hurdles beyond technology. User trust is paramount—people need to feel confident delegating financial decisions to AI agents. This trust will be built gradually through positive experiences in low-stakes scenarios before extending to higher-value transactions.
Education will also be critical. Most consumers and businesses have limited understanding of how AI agents work, what their capabilities and limitations are, and how to set appropriate authorization parameters. Service providers will need to invest heavily in user education and intuitive interfaces that make agent management accessible to non-technical users.
The infrastructure layer—including custody solutions, compliance tools, and dispute resolution mechanisms—needs to mature significantly. Current systems were designed for human-initiated transactions and will require fundamental rethinking to support autonomous agent activity at scale.
Finally, the economic model needs to prove itself. For Agentic Commerce to succeed, it must deliver clear value to all participants—consumers, merchants, and platform providers. Early use cases will need to demonstrate tangible benefits in terms of time saved, costs reduced, or experiences enhanced.
Conclusion: A Transformative Shift in Commercial Infrastructure
The negotiations between Tencent and Manus, alongside Meta's commitment to Agentic Commerce and stablecoin payments, signal a fundamental transformation in how commercial transactions will be conducted. This is not a distant future scenario but an emerging reality that is already reshaping competitive dynamics among tech giants.
For market participants—whether payment providers, financial institutions, merchants, or technology platforms—the imperative is clear: the infrastructure, compliance frameworks, and security mechanisms needed to support AI agent commerce must be developed now. Those who position themselves early in this ecosystem will have significant advantages as adoption accelerates.
Stablecoins' role as the default payment rail for AI agents will only grow more important as the Agentic Economy develops. But as Schultz noted, the challenge is not just technical—it is about bringing regulators, businesses, and users along on this journey. Success will require not just innovation but also patient work to build trust, establish standards, and demonstrate value.
The future of commerce is autonomous, digital, and global. The question is not whether this future will arrive but how quickly and which companies will shape its development. The moves by Tencent and Meta suggest the race is already well underway.
Source: link