This article is reprinted with authorization from Automatic Insight, author: Rhythm Editorial Department, copyright belongs to the original author.

In 1119, shortly after the First Crusade, a French knight named Hugo de Payens and his eight companions founded a new monastic order. Their initial intention was simple: to protect pilgrims traveling from Europe to Jerusalem. At that time, the pilgrimage route spanned thousands of kilometers, fraught with danger, and pilgrims carrying large amounts of gold were akin to moving treasure troves, often becoming targets for robbers.

To address this pain point, the group of knights devised a brilliant solution. Pilgrims could deposit their gold at the order's outposts in London or Paris before setting off, in exchange for a promissory note marked with a password. When they finally arrived in Jerusalem after enduring numerous hardships, they could use this note to withdraw an equivalent amount of currency at the local order's outpost. This monastic order later became famously known as the "Knights Templar."

The Knights Templar inadvertently created the first true multinational bank and payment clearing network in human history.

It did not rely on the physical transportation of gold but was built on a vast trust network that spanned across Europe and the Middle East. A parchment promissory note could be exchanged for real money because it was backed by the order's formidable military power, strict organization, and unassailable reputation.

This ancient system designed for human pilgrims is not fundamentally different from our current payment systems; both are built around a recognizable and trustworthy "person."

However, when the initiator of transactions is no longer a pilgrim heading to the holy city but an AI making decisions in milliseconds in the digital world, the old foundation of trust begins to shake. AI is not human; it cannot provide biometrics, real-name accounts, or bear legal responsibility. When it starts initiating transactions, the trust mechanism that was originally attached to "people" becomes ineffective.

A new question arises: how do we establish a Knights Templar for machines?

In September 2025, this futuristic question entered the view of the highest global decision-making bodies. The Federal Reserve was the first to express its intention to study AI and tokenized payments, followed closely by former SEC Chairman Paul S. Atkins officially naming it the "Agentic Finance" era in Paris.

When an emerging trend gains public validation, it transforms from the blueprint of visionaries into a race that all giants must scramble to seize. From fintech newcomers to traditional card organization empires, all key players have thrown themselves into the fray.

What exactly is Agentic Finance? Why can it become a pivot for reshaping the entire digital business landscape? And why has the ultimate focus of this war fallen on the seemingly ancient link of "payments"?

Although the SEC Chairman referred to it as Agentic Finance, the industry more commonly uses the term Agentic Commerce. Its core lies in the transformation of AI's identity; it is no longer merely an efficiency tool but is gradually evolving into a market entity capable of conducting economic activities independently.

In the past, AI was a tool that completed tasks according to human preset rules. However, after the advent of large language models, AI agents with autonomous planning and API-calling capabilities began to execute complex operations and initiate collaborations in certain scenarios.

A prototype of a "Machine-to-Machine Economy," where transactions and settlements are completed directly between AIs, is beginning to take shape.

However, collaboration does not equate to a closed loop. If AI can only decompose tasks and call services, ultimately still requiring humans to complete payment actions, its role remains at the tool level.

For AI to truly form a commercial system, the key lies in whether it can complete value exchanges, that is, payments.

Agents lacking native payment capabilities struggle to achieve a true closed loop. When AI has planned everything for you but requires you to manually input a password to complete the payment at the last step, it signifies a disruption in the process and means the agent's autonomous capabilities are rendered meaningless. For Agentic Commerce to be established, AI must be able to autonomously complete the final link.

Whoever can provide stable, embedded payment capabilities for this new type of transaction relationship will have the opportunity to become the controller of the next generation of commercial infrastructure and seize the lifeblood of the new commercial order.

This is precisely the source of both anxiety and excitement for payment giants; they see a historic opportunity for a tool to leap from the background to become a new commercial entry point, but they also realize that our existing global payment system is a massive obstacle on the road to this future.

Because its underlying logic is to prevent machines from intervening, rather than allowing machines to participate.

This logic runs through the architecture, standards, and legal mechanisms of payment systems. Taking credit card payments as an example, its strict PCI DSS security standards prohibit AI from accessing sensitive information such as card numbers and expiration dates in any form. At the same time, to prevent automated fraud, verification codes, biometrics, and other anti-robot mechanisms are commonly embedded in payment processes, fundamentally blocking the autonomous operation path of AI.

A deeper contradiction arises from the change in usage scenarios. Traditional payment models are built on human consumption habits, where one order corresponds to one payment; the process, though complex, is controllable.

In AI-dominated scenarios, transactions begin to become high-frequency, massive, small-scale, and continuous. For example, an agent may purchase data, call models, and exchange computing power with multiple servers every second to optimize ad placements. In this environment, fee structures, transaction thresholds, and response delays become insurmountable obstacles for traditional payment models.

Legal systems also lack the capacity to support AI payment behaviors. Taking the "chargeback" mechanism as an example, which is one of the core systems protecting consumer rights, it renders transactions initiated by AI legally lacking immutable settlement efficacy. AI cannot become a clearing entity, and thus cannot truly bear independent economic responsibility.

Jack Forestell, Chief Product and Strategy Officer at Visa, has publicly stated, "Agent-based e-commerce has performed excellently in shopping and discovery phases, but has encountered significant trouble in the payment phase."

This is one of the current development bottlenecks of Agentic Commerce. Despite rapid technological advancements, agents are already capable of completing complex tasks but are stuck at the final step. As long as they cannot cross the payment threshold, AI agents can only remain at the stages of chatting and recommending, unable to truly become a part of the commercial system.

In the past commercial system, payments always played a supportive role, with service platforms being the core that defined transaction paths. However, when transactions are initiated by AI agents, the original primary and secondary relationships are being rearranged. A new structure born around AI's action logic is replacing the old commercial model, pushing payments to the center of the power structure.

To understand the profundity of this power shift, we must first analyze the role of payments in traditional commerce.

The essence of any commercial transaction is "delivery and payment." Among them, the "delivery" of goods or services is the dominant action driving the transaction, defining its nature and purpose.

Meanwhile, "payment," as a means of value settlement, plays a supportive and cooperative role. It is like an ordinary merchant traveling to the Middle East; their primary goal is to "do business" (delivery), while hiring the Knights Templar to protect their gold (payment) is merely a subordinate step to ensure the smooth completion of the business.

Therefore, the fundamental value of traditional third-party payment systems lies in facilitating an already existing "delivery demand," making the "payment" process smoother. Whether it is bank transfers or PayPal, their business is subordinate to a specific commercial scenario defined by service platforms.

However, in the structure of Agentic Commerce, this primary and secondary relationship has fundamentally reversed.

At the top of this new structure is the entry layer directly facing users, where users interact with AI agents (such as OpenAI's ChatGPT, Apple's Siri), forming the traffic entry point of the new era. This layer is responsible for capturing and interpreting users' most original commercial intentions and deciding how to distribute and execute them.

The bottom layer consists of traditional service layers, where past platform giants like Uber, Shopify, and Meituan are categorized. They no longer directly face consumers but passively API-ify their core service capabilities, transforming into "skill plugins" waiting for AI calls, significantly diminishing their brand influence and user interface.

Crucially, the payment layer, positioned in the middle, connects top-level intentions with bottom-level services—this is the vital artery that all giants are vying for. This core hub, composed of PayPal, Stripe, and future crypto wallets, not only controls the final closure of transactions but also becomes the value settlement center and power core of the entire new commercial paradigm.

The core capability of AI agents lies in autonomously and end-to-end completing tasks. To achieve this, a seamlessly integrated, pre-authorized, and highly reliable payment settlement capability is no longer an option for the final task link but is the foundational basis and core capability for AI agents to execute all commercial tasks.

Thus, the payment layer has transformed from a "subordinate function" that cooperates with specific transaction scenarios into the underlying infrastructure that empowers all AI agents to conduct transactions.

This role shift leads to a reversal of commercial logic.

In the past, commercial behavior began with "service." A pilgrim's thought process is that they want to go to Jerusalem (service), so they need the Knights Templar to ensure the safety of their property (payment). Payment exists for the sake of service; it is a supporting step.

But in the world of Agentic Commerce, the logic is inverted. The execution path of AI agents is no longer "first there is service, then seek payment," but rather the opposite; because it has already built in a stable, trustworthy, real-time settlement payment system, it has the capability to autonomously initiate and complete commercial tasks.

Payment is no longer the endpoint of a transaction but the starting point of a new transaction.

The behavioral logic of AI agents is not "I want to use a certain service, then choose a payment method," but rather the reverse; based on the payment capabilities I have already integrated, I look for what services I can directly call to quickly complete transactions.

In their view, Uber is not an app but a transportation function that can be accessed via API. The premise for choosing it is not brand recognition or user habits, but whether it can be efficiently called by the existing payment system.

This means that whoever controls the payment interface controls the priority of service calls. In an AI-dominated world, payment precedes service, becoming the starting point that defines transaction paths.

The key lies in who can seize this starting point.

The problem is clear enough, so the opportunity is equally clear.

Faced with the imminent new wave, payment giants are not remaining in strategic observation but are engaging in an arms race concerning the future dominance of commerce with real money and core product lines. Their paths may differ, but their goals are remarkably consistent—occupying the most critical payment choke point in the new order.

PayPal's New Gang

As the pioneer of global online payments, PayPal's strategy is three-pronged, attempting to build an Agentic Commerce ecosystem around its payment capabilities, positioning itself as an enabler of the new economy.

In April 2025, PayPal launched the Agent Toolkit, a meticulously designed toolkit that allows developers to easily integrate PayPal's core business functions such as payments, shipping tracking, and invoice management into agent programs built on mainstream AI frameworks (like OpenAI's Agent SDK and LangChain), much like LEGO blocks.

Just a month later, PayPal announced a strategic partnership with the rising AI search company Perplexity. This collaboration is highly symbolic, allowing users to complete the entire process from product search and price comparison to final purchase directly within Perplexity's chat interface, using PayPal or Venmo for payment.

In September 2025, PayPal Ventures made consecutive moves, focusing on the underlying capabilities of Agentic Commerce.

It led an investment in the blockchain startup Kite, which is building an identity verification, application distribution, and settlement network for AI agents, attempting to establish a foundational protocol for the machine economy. A few days later, PayPal announced an investment in the stablecoin public chain Stable, integrating its own stablecoin PYUSD and beginning to participate in the on-chain payment ecosystem led by stablecoin giant Tether.

From Kite to Stable, PayPal's two investments revolve around the same direction. When transactions are initiated by AI, the existing identity, interaction, and payment systems will struggle to sustain. PayPal is preemptively positioning itself for this new system, ensuring it remains at the core of future value circulation networks.

Stripe's Technological Moves

If PayPal's strategy is to embrace and empower the existing ecosystem, then Stripe's ambition is even grander; it seeks to reconstruct the payment tracks of the AI era from the ground up, becoming a rule-maker.

In May 2025, Stripe released its Payments Foundation Model, a transformer model trained on its vast transaction data, specifically designed to optimize fraud detection and payment authorization rates. Stripe is using AI technology to reconstruct its core risk control system, transforming experience-based judgments into standardized capabilities.

Stripe's trump card was revealed in September 2025 when it announced a partnership with Paradigm, one of the top venture capital firms in the cryptocurrency space, to incubate a new L1 blockchain called Tempo. Tempo's positioning is extremely clear: it is born for payments, aiming to provide a high-throughput, low-latency, and low-cost underlying infrastructure for stablecoins and real-world payment scenarios.

From the design phase, Tempo has brought in first-line participants from the AI, e-commerce, and financial industries as advisors, including OpenAI, Anthropic, Shopify, and Visa. Stripe's intention is clear: not only to serve Agentic Commerce but also to participate in setting its technological order.

The Old Strongholds of Visa and Mastercard

In the face of the aggressive advances of PayPal and Stripe, the two traditional card organizations, Visa and Mastercard, have not remained idle. Their core strategy is to protect their vast and profitable existing networks while striving to position themselves as indispensable trust intermediaries in the new commercial paradigm.

Both companies' Agentic payment plans, whether Visa's Intelligent Commerce platform or Mastercard's Mastercard Agent Pay, focus on several key points. First is agent identity verification, ensuring that only authorized and registered AI agents can initiate transactions.

Second is payment tokenization, which replaces real card information with a one-time secure token for transactions. This is a security technology they have promoted for years, which is particularly important in AI agent payment scenarios, as it provides AI with a usable payment credential without touching the PCI DSS red line.

Finally, there is consumer authorization and control, ensuring that every transaction initiated by an agent has the explicit consent of the user and can be revoked at any time through technologies like biometrics.

Their actions are based on clear market data.

According to Mastercard's data, the number of active users of OpenAI doubled in just a few weeks in 2025, reaching 800 million. Meanwhile, Agentic AI is expected to handle up to 20% of e-commerce tasks in 2025. As consumer trust, AI capabilities, and merchant adoption rates increase, analysts predict that the market size will grow to $1.7 trillion by 2030, with a compound annual growth rate of 67%.

The Agreement Dream of Coinbase and Google

While traditional payment giants are working to transform existing systems to adapt to the AI era, a more native payment infrastructure blueprint tailored for the machine economy is being jointly drawn by the crypto world and tech giants.

The disruptive nature of the crypto world’s solution lies in its provision of three native capabilities to AI agents that traditional finance cannot offer: permissionless economic identity (on-chain wallets), immutable final settlement (resolving transaction disputes), and programmable economic contracts secured by smart contracts. Guided by this philosophy, Coinbase's exploration is particularly aggressive.

After empowering over 20,000 AI agents with on-chain wallet digital identities through AgentKit, Coinbase quickly launched the x402 protocol, aimed at solving the value exchange dilemma between AI and internet services.

The x402 protocol activates the long-dormant HTTP 402 "Payment Required" status code, natively coupling payments with API service calls at the protocol level. When an AI agent accesses a service, it receives a "bill" challenge containing payment information and can immediately settle on-chain using stablecoins to gain service access.

If x402 focuses more on the efficiency of transactions between machines, then the AP2 (Agent Payments Protocol) released by Google in collaboration with over 60 payment networks, financial institutions, and tech companies aims to address the fundamental issues of trust and compliance in AI payments.

When the initiator of a transaction shifts from identifiable humans to AI agents, how to confirm the user's true intent and define responsibility becomes a question that the entire commercial system must answer.

The emergence of AP2 is intended to provide a regulatory common syntax for how AI spends money. Google is adopting an open-source strategy, attempting to establish AP2 as a de facto industry standard before the market landscape solidifies.

Google's ultimate goal is to become an integrator of the next generation of financial services. Just as NFC chips embedded authentication capabilities into mobile hardware, AP2 aims to embed the ability to call financial services in a standardized way into future AI agents. If successful, Google will become the master switch connecting everything in the Agentic Commerce ecosystem.

In summary, different forces are starting to compete from scratch on the same track.

In the short term, PayPal's actions are the most direct. Through the Agent Toolkit, it quickly embeds payment capabilities into AI application development; its collaboration with Perplexity has turned "conversational transactions" from a concept into a verifiable reality; at the same time, it has invested in Kite and Stable, introducing its own stablecoin PYUSD, and has taken the lead in laying out the on-chain payment ecosystem.

In the medium term, Stripe's path leans more towards reconstruction. If the Tempo public chain can operate stably, it will provide a new track for payment infrastructure, elevating itself from a participant to a framework maker; combined with the potential of the Payments Foundation Model in risk control, Stripe is expected to form a differentiated advantage in the new landscape.

From a long-term perspective, the coupling of crypto-native protocols and the machine economy is even higher. They provide native settlement paths for high-frequency automated microtransactions between AI-2-AI, and once scaled and matured, could reshape the overall market structure. Meanwhile, Visa and Mastercard, relying on their merchant networks and trust systems, remain the core barriers in the existing market.

Overall, in the short term, we look at PayPal's speed of implementation; in the medium term, we observe Stripe's structural attempts; and in the long term, we assess the compatibility of crypto-native solutions. Different paths, yet pointing towards the same consensus: controlling payments in the AI era means controlling the future of commerce.

Agentic Commerce brings not only a reshuffling of power dynamics but also profound impacts on all participants in the business world, raising a series of new questions that need to be answered.

In the traditional internet era, Google defined the priority of information through its search algorithms, spawning a massive search engine optimization industry. Merchants paid Google to have their businesses prioritized on search pages, which is Google's core revenue model.

Now, as the search entry point shifts from Google's search box to ChatGPT's dialogue box, a new question arises: who defines the recommendation logic of AI agents? How can merchants achieve high rankings in agent recommendations? In the future, will there also be paid rankings, allowing the highest bidding merchants to be prioritized by agents?

This transformation has also attracted significant attention from global regulatory agencies. In September of this year, the Federal Reserve announced it would specifically discuss "emerging stablecoin use cases" and "the tokenization of financial services," clearly stating that it is studying the applications of AI and tokenization in the payment field.

Regulatory intervention may pave the way for innovation but could also bring new compliance challenges. Whoever can propose a set of payment standards that meet regulatory requirements while satisfying the needs of Agentic Commerce may become the de facto industry leader.

The era of Agentic Commerce has already begun. It promises users an unprecedented minimalist experience but also pushes traditional platforms to the edge of being API-ified. In this profound structural transformation, the payment layer is leaping from a behind-the-scenes auxiliary role to becoming the vital artery controlling the lifeblood of commerce.

The fierce positioning battle among payment giants signals that a major upheaval in the business world is imminent. In this upheaval, whoever can first secure the closed loop of payments will grasp the throat of the new commercial era. This is not only a competition of technology and business models but also a war over the future ownership of the central bank's scepter in the digital world.

Related: Opinion: The mass adoption of Web3 requires embracing Web2 infrastructure rather than replacing it.

Original: “The AI Payment Era: The Reshuffle of Payment Giants”

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。