The Battle for the AI Payment Race: Traditional Card Networks Face Off Against Coinbase

marsbitPublished on 2026-06-08Last updated on 2026-06-08

Abstract

With the rise of AI agents conducting transactions, a battle for the underlying payment infrastructure is underway. Two distinct and incompatible approaches have emerged for enabling autonomous AI payments. The first approach is championed by traditional card networks Visa and Mastercard. They leverage their existing tokenized card credential systems, extending them to allow verified AI agents to make purchases within user-defined limits. Services like Mastercard's Agent Pay and Visa's Intelligent Commerce integrate with major AI platforms (e.g., OpenAI, Anthropic) and keep transactions within the established, decades-old card payment model. This system offers advantages for consumer retail, including robust fraud protection, chargeback mechanisms, and extensive merchant networks. The second approach, led by Coinbase, utilizes stablecoins on open internet protocols. Its x402 protocol reactivates the HTTP 402 status code for machine-to-machine micropayments, using USDC for settlement directly on-chain. This method eliminates the need for accounts or card fees, making it highly efficient for high-frequency, low-value, cross-border transactions between AI agents—such as paying for API calls, data streams, or computational resources—where traditional card fees and settlement times are impractical. While card networks excel in consumer-facing scenarios requiring dispute resolution, stablecoin protocols are tailored for machine economies. A key challenge for both is agent identi...

Written by: Zennon Kapron

Compiled by: Chopper, Foresight News

As AI agents increasingly undertake various commercial transactions, a battle for the underlying payment infrastructure has already begun.

Currently, the technical path to enabling AI agents to make autonomous purchases has diverged into two incompatible solutions: through which channel transactions are ultimately cleared and settled when software programs make payments on behalf of users. One camp builds payment rails based on tokenized bank card credentials controlled by Visa and Mastercard; the other, led by Coinbase, adopts stablecoins for settlement based on open internet protocols. While the surface-level focus of AI agent commerce is on shopping assistant applications, the core game behind the scenes is about who will dominate the next-generation payment system.

Two Payment Channels, Adapted to Different Application Scenarios

Traditional card networks have taken the initiative and moved swiftly. Mastercard launched its Agent Pay service in April 2025, built on its proprietary agent tokenization system. This tokenization technology, originally designed for contactless payments and card-on-file payment scenarios, has been extended to allow verified AI agents to make transactions on behalf of users within authorized limits.

From its launch, the service assembled a host of industry partners, signaling a clear strategic intent: collaborators include Microsoft, IBM's watsonx intelligent orchestration platform, and payment service providers Braintree and Checkout.com. A day later, Visa introduced its Visa Intelligent Commerce service, opening its payment network to AI developers, with AI-adapted bank cards as the core vehicle. This solution replaces raw card numbers with tokenized credentials, proving user authorization for a specific AI agent and defining transaction boundaries. Visa also enlisted several leading AI companies, including Anthropic, OpenAI, Perplexity, Mistral, and Samsung.

The solutions from both major card networks keep transactions within the decades-old bank card payment model. While AI agents are a new entity, the payments flow through traditional channels that have served global commerce for half a century.

The stablecoin camp adopts an architecture fundamentally different from the card networks. Coinbase launched the x402 protocol in May 2025, reviving the long-dormant HTTP 402 "Payment Required" status code. The protocol leverages this code to settle transactions using the USDC stablecoin directly on the internet. The specific process involves: a client requesting access to a resource, the server returning a payment instruction; the client attaching signed stablecoin payment information in the request header; once the on-chain transaction is confirmed, the corresponding resource becomes accessible. The entire process requires no account registration, card binding, and incurs no bank card transaction fees.

This solution is designed specifically for machine-to-machine (M2M) transactions. An AI agent might need to make thousands of micro-payments for API calls, data streams, or interactions with other agents. Processing such transactions through traditional card channels is completely unfeasible from a cost perspective.

Both technical routes have their strengths. The bank card channel excels in personal retail consumption scenarios, which place high demands on chargeback handling, fraud prevention, and dispute arbitration mechanisms. The stablecoin channel holds significant advantages for high-frequency, small-value, cross-border machine transactions, where traditional card fee structures and settlement times are ill-suited. The core of the contest lies in which type of scenario will become the mainstream for AI agent commercial transactions.

A major challenge common to both routes is identity verification. When a software program initiates a payment, merchants need to confirm that the operator is a legitimate agent authorized by a real user, not a malicious bot using stolen credentials. Simultaneously, users need a mechanism to dispute transactions erroneously initiated by an AI agent.

Visa, noting a 47-fold surge in AI traffic to US retail websites, collaborated with cloud service provider Cloudflare to launch a Trusted Agent Protocol for distinguishing legitimate AI programs from malicious crawlers. This also highlights the structural advantage of traditional card networks: five decades of accumulated risk scoring systems, chargeback rules, and dispute resolution mechanisms are well-suited to handle issues like AI agents purchasing the wrong item. In contrast, stablecoin transactions are immutable once on-chain, with no native reversal solution within the system.

Looking ahead, the key factor determining the direction of the consumer market may not be which payment channel has lower fees, but who can solve the challenges of agent identity verification and transaction dispute resolution.

Dual-Track Strategy: Card Networks Betting on Both Arenas

A telling signal is that Visa and Mastercard are not betting everything on their own rails but are also simultaneously investing in the stablecoin arena.

By April 2026, Visa's stablecoin settlement business had reached an annualized transaction volume of $7 billion, a 50% increase quarter-over-quarter. The company added support for 5 new public blockchains, bringing its total number of partner chains to 9, while launching over 130 "stablecoin + bank card" linkage projects in more than 50 countries globally. In October 2025, Visa doubled down, jointly launching the Trusted Agent Protocol with Cloudflare to help merchants identify legitimate agents versus malicious programs. It also publicly announced a collaboration with Coinbase to promote interoperability between its network and the x402 protocol. Seemingly competing bank card systems and stablecoin protocols are now building bridges.

Mastercard has adopted a similar dual-track strategy. In March 2026, Mastercard announced its intention to acquire stablecoin platform BVNK for up to $1.8 billion. Prior to this, its Agent Pay service had expanded to Latin America and the Caribbean, with local issuer adaptation completed in early 2026.

It's clear that the core strategy of the two major traditional card networks is: no longer merely defending the bank card channel, but striving to become the fee-collecting gateway for all payment rails, whether proprietary or stablecoin-based. This layout strongly indicates their assessment: if the industry ultimately settles on bank cards as the mainstream for AI payments, they would not need to invest heavily in acquiring stablecoin infrastructure.

Diverging Application Scenarios

Based on currently launched products, the application boundaries of the two technical routes are quite clear.

Mainstream products targeting ordinary consumers mostly choose the bank card channel. The "Checkout with ChatGPT" feature launched in September 2025, co-developed by OpenAI and payment service provider Stripe, relies on a shared payment token for bank card settlement. This token is limited to specified merchants and shopping orders, initially connecting with Etsy sellers and later covering over a million Shopify stores. Amazon's "Buy for Me" feature, which allows AI agents to make purchases on third-party websites on a user's behalf, automatically populates the user's stored bank card for checkout.

AI shopping services for personal consumption commonly choose bank cards due to the system's mature anti-fraud tools, extensive merchant network, and long-established user trust.

Meanwhile, the stablecoin channel firmly occupies the machine transaction market. Amazon integrated the x402 protocol into its Bedrock agent core payment service, using Coinbase's Base blockchain for settlement, with transactions taking about 200 milliseconds and fees under one cent. Stripe also joined the service as a payment gateway. According to Coinbase data, in its first year, the x402 protocol processed over 169 million payment orders, covering 590,000 buyers and 100,000 sellers.

These transactions are not about users shopping for clothes online, but AI agents paying for services like computing power, data, and API calls. Their frequency and individual amounts are incompatible with the design logic of bank cards. In September 2025, Coinbase, jointly with Cloudflare, spearheaded the establishment of the x402 Foundation, aiming to promote the development of universal industry standards rather than building a closed proprietary product.

Summarizing five flagship AI commercial payment projects launched by early 2026: 3 used bank card settlement, 2 used stablecoin settlement, with application scenarios largely divided along the lines of personal consumption and machine transactions.

Future Industry Direction

In the short term, the industry landscape in 2026 is likely to maintain the status quo: bank cards dominating personal retail payments, stablecoins specializing in inter-machine transactions, with both coexisting and developing. However, by 2030, this situation may be disrupted, as both camps are vying for the convergence zone of the two scenarios.

The ultimate decider will depend on whether AI-driven commercial transactions ultimately lean more towards traditional retail forms or evolve into a massive network of micro machine transactions. If it's the former, traditional card networks will retain dominance. If it's the latter, stablecoin channels will capture a significant volume of new transaction flows.

Visa and Mastercard have made the safest choice: a dual-track investment in both arenas, ensuring they can collect fees regardless of where future flows go. The entities that truly need to be cautious are those betting on a single payment channel. The two major card networks have already hedged this risk, which intuitively reflects their judgment about the industry's future.

Trending Cryptos

Related Questions

QWhat are the two main technical approaches for enabling AI agents to conduct autonomous payments, as discussed in the article?

AThe two main approaches are: 1) The traditional card network model led by Visa and Mastercard, which uses tokenized card credentials for settlement. 2) The stablecoin-based model led by Coinbase, which uses the x402 protocol and USDC stablecoin for settlement directly over the internet.

QAccording to the article, what is a key structural advantage that traditional card networks like Visa possess in the AI payment space?

AA key structural advantage is their 50-year history of developing risk scoring systems, chargeback rules, and dispute resolution mechanisms. This existing infrastructure is well-suited for handling issues like incorrect purchases made by AI agents, which are difficult to address in native, irreversible stablecoin transactions.

QWhy do personal consumer AI shopping services typically prefer the card network payment channel?

APersonal consumer AI shopping services prefer the card network channel because it offers mature anti-fraud tools, an extensive merchant network, and long-established user trust. This makes it more suitable for retail purchases where issues like refunds and disputes are common.

QWhat is the strategic rationale behind Visa and Mastercard's 'dual-track' approach of investing in both their traditional networks and the stablecoin sector?

AThe strategic rationale is to ensure they remain the fee-collecting gateway for all payment flows, regardless of which channel (traditional card or stablecoin) ultimately dominates AI agent commerce. This approach hedges their bets and shows they are not relying on a single payment future.

QWhat is the primary use case for the stablecoin-based x402 protocol, and what specific transaction characteristics make it suitable?

AThe primary use case for the x402 protocol is machine-to-machine (M2M) transactions. It is particularly suited for high-frequency, low-value, cross-border micropayments, such as payments for API calls, data streams, or interactions between AI agents, where the traditional card network's fee structure and settlement speed are impractical.

Related Reads

7 Months After the Collapse of Huiwang, Southeast Asia's Escrow Platforms Undergo a Major Reshuffle

Following the collapse of Huione Pay—dubbed the "Alipay of Southeast Asia"—seven months ago, the region's underground financial guarantee platform sector is undergoing a significant reshuffle. This power vacuum has been swiftly filled by emerging platforms such as XinBi, Tiger/Navigator, JinBei (renamed JinBo), Dali/Tiancheng, and FullyLight. These platforms, operating largely via Telegram and offering services like escrow for illicit transactions, have absorbed the vast user base and markets left behind by Huione. While positioning themselves as "trust intermediaries," their primary clientele consists of networks involved in online scams, money laundering, illegal gambling, and even human trafficking. For instance, the Tiger/Navigator platform explicitly provides "escrow" services for kidnapping-for-ransom operations ("强押车交易"). Data underscores the immense scale: Huione alone processed over $103 billion in cryptocurrency payments and facilitated over $31 billion through its escrow market before its downfall, linking it to Cambodia's notorious Prince Group. Since its collapse, competitors have seen explosive growth. For example, the XinBi platform has accumulated over $1.6 billion in total USDT revenue, while platforms like NewPay, OkPay (under Dali), and FullyLight Wallet collectively processed over $4.8 billion in USDT in a single year. This ecosystem thrives in regions like Cambodia and Myanmar, where regulatory gaps allow these platforms to act as critical financial infrastructure for sprawling cybercrime industries, from scam compounds to online casinos. The article concludes that the moniker "Southeast Asian Alipay" is a misnomer, obscuring the platforms' fundamental role in enabling serious criminal enterprises rather than representing legitimate financial innovation.

Odaily星球日报1h ago

7 Months After the Collapse of Huiwang, Southeast Asia's Escrow Platforms Undergo a Major Reshuffle

Odaily星球日报1h ago

The Changing Landscape: What Are Crypto VCs Experiencing?

Title: The Shifting Landscape of Crypto Venture Capital The era of dedicated crypto venture capital funds is undergoing a significant transformation. Once essential for navigating the sector's complexity and high risk, these specialized funds are now facing an identity crisis as the market matures. This shift mirrors historical patterns in other specialized investment classes like cleantech and SPACs, where initial information advantages dissipate as technologies become mainstream and integrated into existing industry frameworks. The article argues that crypto is reaching a critical inflection point, transitioning from a "building phase" to an "integration phase." Major players like Stripe, BlackRock, and Visa now engage with crypto not for its novel mechanics but as a foundational financial infrastructure. Their needs—regulatory compliance, banking partnerships, distribution channels—align with traditional fintech, a domain easily understood by large, generalist funds like Sequoia and Founders Fund. This evolution creates a "barbell effect" within the VC landscape. On one end are massive, diversified platforms that can incorporate crypto as one vertical among many. On the other are small, nimble funds focused on niche, experimental projects. The middle ground—medium-sized dedicated crypto funds—is being squeezed out. Their typical fund size makes it impossible to generate sufficient returns solely from early-stage crypto bets, yet they cannot compete with giants for later-stage deals. Consequently, leading crypto-native firms like Paradigm and Framework Ventures are expanding into AI, robotics, and other sectors, driven partly by LP pressure for better returns amid a broader VC DPI crisis. Others, like Dragonfly and a16z, have narrowed their crypto focus predominantly to financial infrastructure like stablecoins, reframing the sector's core narrative. For crypto entrepreneurs, this consolidation presents challenges. While generalist funds offer larger checks and broader resources, crypto projects now compete fiercely with AI for attention and capital within these firms. Furthermore, the long-term, non-commercial foundational work that built the ecosystem—funded by dedicated crypto VCs—is less likely to attract generalist capital focused on direct returns. The conclusion is that "crypto investor" as a standalone category is becoming obsolete, akin to "internet investor." Crypto is becoming a baseline infrastructure layer. The future will see a barbell structure: large-scale growth financing handled by generalist funds, while pioneering, speculative projects are funded by small, specialized vehicles. The dedicated crypto funds of the 2017-2021 boom, which incubated core infrastructure, are giving way to this new, bifurcated reality.

Foresight News1h ago

The Changing Landscape: What Are Crypto VCs Experiencing?

Foresight News1h ago

As Consensus Accelerates, What Are Young Investors Betting On?

Title: As Consensus Forms Faster, What Are Young Investors Betting On? In the rapid evolution of tech investment, a new generation of young investors is navigating a landscape where AI, robotics, commercial aerospace, and quantum computing are advancing simultaneously. Traditional investment logic based on financial models is giving way to a need for deep technical understanding and the ability to act before industry consensus forms. An analysis of trends from the "WAIC FUTURE TECH" list of young investment leaders reveals key shifts in focus. The first major trend is the movement of AI from the digital screen into the physical world. Investment is shifting from large language models and chatbots towards embodied AI, robotics, AI hardware, and edge computing. While demonstrations generate excitement, the real challenge lies in achieving scalable, reliable, and cost-effective delivery in complex real-world environments like factories and logistics. Success depends not just on algorithms but on the integration of sensors, actuators, and control systems. Second, the competitive focus for large models is moving beyond raw capability toward building an "intelligence flywheel." The goal is to create self-reinforcing systems where user interaction generates data, improving the model, which in turn enhances the user experience and attracts more engagement. Companies that successfully embed AI into workflows to create these closed-loop systems can build lasting value that isn't easily erased by the next model upgrade. Third, facing a potential bottleneck in high-quality human-generated data, investors are looking at new underlying technologies. Reinforcement learning and self-play, as demonstrated by AlphaGo Zero, offer paths for AI to generate its own experience. Scientific foundation models, which aim to build general AI capabilities for fields like life sciences and materials discovery, represent a non-consensus direction that could unlock new frontiers of knowledge and data. Finally, in deep-tech areas like quantum computing, commercial aerospace, and space-based infrastructure, patient capital is essential. These fields have long, uncertain development and validation cycles involving complex engineering, supply chains, and regulations. Investment here requires a long-term view, focusing on foundational team capabilities and the eventual emergence of market demand, even if commercial returns are distant. Collectively, these trends illustrate how young investors are adapting to a new era. They are learning to make earlier, technically-informed judgments, balance hype with real-world viability, and provide the patient capital needed to build the deep-tech foundations of the future.

marsbit1h ago

As Consensus Accelerates, What Are Young Investors Betting On?

marsbit1h ago

Trading

Spot

Hot Articles

Discussions

Welcome to the HTX Community. Here, you can stay informed about the latest platform developments and gain access to professional market insights. Users' opinions on the price of AI (AI) are presented below.

活动图片