Agentic Payments 2026: How AI Agents Are Reshaping Commerce and Payment Infrastructure

Just as people, AI agents have much more possibilities to fulfil their potential if they have money to back their actions. Without the ability to transact, all the search and analysis agentic systems perform remain on theoretical level. To empower AI agents with more practical tasks, they need agentic payment mechanisms that are already emerging and rapidly scaling. This article explores protocols supporting agentic commerce, infrastructure, legislative and public readiness to the concept of AI-assisted shopping and agentic AI payments market in 2026.

In September 2025, OpenAI launched a novel but logical addition to its popular chatbot – ChatGPT Instant Checkout. For the first time, a mainstream AI assistant could not just recommend a product, it could buy the recommended goods directly. Although in March 2026, the AI company decided to scale back the direct checkout plans and focus on agentic product discovery, while routing purchases through third-party apps and retailer platforms, the idea of AI-driven commerce set firmly in consumer minds. In fintech industry circles, the idea has been nurtured for years now. Yet, with the advent of supporting infrastructure, in 2026 the agentic payments and commerce market finally promises to show the first signs of blooming.

Agentic Commerce Share in AI Market Size and Growth Projections

Agentic commerce, i.e. commercial activities involving purchase transactions initiated and/or completed by AI agents acting on behalf of humans, is the fastest-growing subsector of the broader agentic AI market.

The agentic AI market reached $5.2 billion in 2024 according to Market.us Research, and is projected to expand to $196.6 billion by 2034 — a compound annual growth rate of 43.8%. However, the size of e-commerce transactions that may be performed by AI agents is much bigger.

Juniper Research estimates total agentic commerce transaction value in 2026 as $8B and estimates that its value may be growing at a total cumulative growth rate of 43,240% throughout 2031, reaching $3.5 trillion in 2031. It coinsides with other industry projections as well. McKinsey projects that agentic commerce alone could facilitate $3 to $5 trillion in consumer sales globally by 2030. Notably, this impressive number is only a moderate growth scenario.

If we compare the estimated 2030 agentic AI market size of $46B, and forecasts for agentic commerce transaction value, which imply about 237% CAGR by our calculations, we can see that agentic commerce payments will reach about $1.04T in 2030, surpassing agentic AI market size by about 2,161%. The two trajectories are shown in Chart 1 below.

Chart 1: Agentic AI Market Size vs Agentic Commerce Transaction Value Comparison

At the same time, the actual compound annual growth rate for agentic commerce is currently hard to estimate since we’re only at the starting point of this technology, and there’s little known about the growth of infrastructure and legislation that is supposed to underpin agentic commerce growth.

How Agentic Payments Work: The Basics

On the surface, agentic payments function just as traditional user-made ones except for that the payer is actually an AI agent instructed by users. However, accomplishing these agent-led transactions requires different mechanisms since confirming the transaction intent in the autonomous environment is different than authenticating a personal purchase made by a human.

To enable this function, the new protocol layer sits between AI models and payment infrastructure. In 2026, several primary standards compete for agentic payment protocol dominance:

Agentic Commerce Protocol (ACP)

• Developed jointly by OpenAI and Stripe, live since September 2025 in ChatGPT;
• Processes payment via Stripe’s standard merchant acquiring infrastructure;
• Charges a 4% OpenAI platform fee + about 3% Stripe processing fee = 7% in total for autonomous agent-led conversions.

Universal Commerce Protocol (UCP)

• Announced by Google at NRF 2026 (January 11, 2026) and described as an open standard;
• Co-developed with Shopify, Etsy, Wayfair, Target, and Walmart;
• Enables Native Checkout within Google Search’s AI Mode and Gemini;
• Endorsed by Visa, Mastercard, American Express, Stripe, and Adyen;
• Connects via APIs, Agent2Agent (A2A), and Model Context Protocol (MCP) for AI agent data access;
• Along with UCP, Google also has Agent Payments Protocol (AP2) that functions as a dedicated trust layer for secure transactions.

Agentic Payment Protocols by Visa & Mastercard

Beyond ACP and UCP, major payment networks Visa and Mastercard have launched their own agent-specific frameworks. Visa’s Intelligent Commerce initiative (April 2025) provides APIs and SDKs for tokenization, authentication, and transaction controls specifically designed for AI agents. Partners included Anthropic, IBM, Microsoft, Mistral AI, OpenAI, Perplexity, Samsung, and Stripe.

Mastercard’s Agent Suite, launched in Q2 2026, focuses on embedding intelligent product discovery for banks and blending personalization with conversational shopping for merchants. Along with Mastercard’s Verifiable Intent offering, that implements Know Your Agent (KYA) frameworks — cryptographic verification systems to distinguish legitimate shopping agents from malicious bots, Agent Suite is a solid platform for AI agents performing online transactions.

Blockchain-Based Payment Protocols x402 and ERC‑8004

Blockchain is one of the most natural financial environments fit for agentic transactions. It is no wonder, that blockchain-based firms are also developing their agentic commerce solutions.

Coinbase-supported x402

x402 is a protocol that integrates payments directly into the fabric of the internet. It repurposes the seldom-used HTTP 402 “Payment Required” status to allow real-time payment requests as part of regular web interactions. Besides enabling individuals agent transactions, x402 has recently introduced batch settlement, a new payment scheme designed for high-frequency, low-value transactions performed by AI agents within predefined rules.

Ethereum-based ERC‑8004

Built on Ethereum, ERC‑8004 is a blockchain standard that creates identity and reputation systems for AI agents.

It introduces three main components:

1. Identity Registry: Each AI agent receives a unique, verifiable identifier on the blockchain.
2. Reputation Registry: Agents accumulate feedback and transaction history, creating a transparent track record.
3. Validation Registry: Confirms that tasks are completed correctly before payments are finalized or reputation is updated.

Chart 2: How Agentic Payments Work

Consumer Trust and Adoption Statistics

One of the biggest obstacles on the way to agentic payments scaling is not even a tech component, but consumer trust. On one hand, people are enthusiastic about delegating their routine shopping search and purchases whether in personal or business environment to an automated mechanism, but they’re also cautious about the outcome of such delegation. Market data better illustrate the scale of public discretion when it comes to AI agents in e-commerce. 

• Only 5% of people globally say they have no concerns at all about agentic commerce;
• Traffic to retail sites from generative AI platforms increased 4,700% year-over-year;
• ChatGPT’s traffic to retail is driving roughly 2x as many new customers as traditional search channels;
• 43% of consumers have had positive experiences with AI-assisted shopping; yet
• 55% worry about identity theft;
• 51% feel they could lose control over their finances;
• 45% of shoppers use AI assistants for product discovery in some form;
• 29% of UK consumers trust AI for automated payments;
• only 16% of US consumers currently trust AI to make payments on their behalf;
• 14% of AI assistant users in Europe simply follow the AI’s top recommendation when making the ultimate purchase decision;
• 92% pay attention to customer reviews when deciding between AI-generated options;
• Walmart saw 3x lower conversion rates for in-chat purchases vs. redirecting users to its own website — a finding that led the retailer to pivot its strategy toward integration rather than replacement.

Giving the positive dynamics of AI agents use in product discovery and similar functions, 40% of enterprise applications will embed AI agents by end of 2026, according to Gartner projections. Nevertheless, users are not entirely wrong about having precautions regarding the AI agents in online shopping settings, since the shift to agentic commerce introduces novel fraud vectors that legacy systems were not designed to address:

• 78% of financial institutions expect fraud to increase significantly as AI shopping agents proliferate;
• when a transaction is triggered via an agent platform, it becomes less clear who is responsible for any dispute arising: the merchant, the agent provider, or the payment processor;
• AI agents exhibit behavior patterns that traditional fraud detection flags as suspicious: rapid sequential orders, purchases across unrelated categories, unusual velocity patterns;

“The fraud systems most merchants rely on today were built to detect bad human behaviour. They were not designed for a world where a legitimate AI agent and a malicious bot look almost identical,” said Monica Eaton, Founder and CEO of Chargebacks911. “As agentic commerce scales, merchants face a clear choice: adapt their detection and evidence infrastructure now, or watch a growing share of legitimate revenue get declined by their own systems.”

To address all those issues, Cloudflare and others are developing cryptographic verification standards (Signature-Input and Signature-Agent headers) to authenticate legitimate agents. Meanwhile, Visa and Mastercard’s Know Your Agent (KYA) frameworks are the industry’s primary response to bot-versus-agent disambiguation.

“The industry has rightly focused on what happens when an AI agent makes a purchase the customer did not want. The question that remains largely unasked is what happens when a merchant’s fraud system refuses the purchase the customer did want,” said Eaton. “Both problems need solving, and both require the same thing: a clear, auditable record of what was authorised and what happened.”

Key Agentic Payment Players and Supporting Infrastructure

The agentic payments ecosystem in 2026 is a four-layer stack: AI platforms that host agents, payment networks that authenticate and settle, commerce platforms that enable agent-compatible storefronts, and infrastructure providers that handle identity, compliance, and rails. Here are some of the key players of this ecosystem:

Chart 4: Key Agentic Payment Ecosystem Players

*Note: The stand-alone Amazon Rufus chatbot will soon be discontinued, but Amazon said it will use Rufus’ recommendation features and shopping history. Now, Amazon brings together Rufus and Alexa+ to create ‘Alexa for Shopping’ on the Amazon Shopping app and website, but, for convenience, we will call the platform Rufus until it fully transforms into the new offering.

What This Means for Payment Service Providers

The rise of agentic commerce restructures the payments value chain in several specific ways that PSPs and payment orchestration platforms must address.

Payment Orchestration Needs a Boost in AI Agent-Driven E-Commerce Environment

At present, payment orchestration is driven by AI but still tends to rely on predefined logic and centralized routing rules. At the same time, the evolution of payment orchestration frameworks and the emergence of agentic AI solutions designed for public use suggest that payment orchestration should eventually move toward agentic payments intelligence as well.

“Agentic orchestration is the most useful for intelligent payment routing, where decisions can be made taking into account vast arrays of information rather than a few predefined rules. Payment orchestration with multiple AI agents (e.g. risk&fraud agent, customer preference agent, success probability agent, each contributing specialized intelligence) allows the system to dynamically evaluate far more variables in real time, as the orchestration agent balances all these trade-offs to maximize success rate, compliance, cost efficiency, and customer experience. […]

agentic orchestration capabilities can serve payment companies while dealing with anti-fraud measures, including related planning and scoring systems, user verification, and compliance.

As for the prospective emerging technologies, we see network tokens actively used today to boost transaction security. Today, token vaults are mostly static repositories, often poorly synchronized, and lacking automated, real-time token lifecycle management. Thus, token vault storage could also be enhanced by agentic AI tools, creating active, adaptive, and intelligent infrastructure optimized for security, speed, and approval success.”

Andrew Riabchuk, Founder of Akurateco

When transactions are performed by AI agents with speed and scale that outperform the payments we have known before, implementing an agentic system into payment orchestration logic is only a natural step forward.

Authorization and Tokenization at Agent Scale

ACP, as an example of emerging agentic payments protocol, uses standard merchant acquiring infrastructure, but at volumes and velocities that differ fundamentally from human-initiated transactions. Tokenization becomes critical, as the effectiveness of tokenized, single-use payment credentials is exactly the model that agent payments rely on.

One of the main problems for payments with AI agents involved is that compliance infrastructure was built for a world of human-paced financial activity. Autonomous AI agents are executing thousands of transactions across multiple platforms in seconds. As AI systems increasingly generate, route, and optimize financial flows, many monitoring tools are struggling to maintain visibility in real time and adopt existing systems to AI demands.

“What is emerging is not just a wave of new software capability, but a structural shift in how financial institutions design and run core processes. Functions such as underwriting support, KYC, fraud detection, treasury operations and regulatory reporting are increasingly being augmented by advanced AI systems, embedding intelligence directly into the workflow layer of banking. This shift raises an important point of balance. While AI can significantly enhance speed, accuracy and efficiency, financial services remain fundamentally built on trust, accountability and controlled execution. That means the real challenge is not adoption, but integration – ensuring these technologies operate within clearly governed, secure and auditable environments.”

Radi El Haj, CEO of RS2

While traditional finance (TradFi) faces pressures from AI-led commercial transactions, particularly in algorithmic trading, fraud detection, and compliance alert overload, for crypto systems, operating on near-instant settlement rails with limited intermediaries, the outcomes of overload are more severe and irreversible.

New Merchant Onboarding Requirements

Merchants who are not agent-compatible will be less visible or harder to recommend by AI shopping systems. Being visible to agents requires:

• Structured, machine-readable product data exposed via APIs;
• Agentic protocols, e.g. ACP and/or UCP, integration;
• Consistent pricing and inventory data that AI agents can trust;
• Agent-compatible checkout infrastructure.

Regulatory Landscape Evolves Slower Than Agentic Payments But There’s a Strong Foundation

Agentic AI in payments is developing faster than regulatory frameworks can follow, but several key developments are shaping the compliance environment:

• EU AI Act: Applies to agentic AI systems, requiring risk assessments and transparency for autonomous systems that make decisions affecting consumers;
• EU Digital Identity Wallet: The EU is introducing digital ID wallets in 2026 for user authentication — a foundational piece for digital identity verification, which might be helpful to distinguish between humans and bots in digital commerce settings; 
• India Aadhaar: Powers over 2 billion monthly authentications as of 2025, providing a model for agent-compatible identity infrastructure at scale. It demonstrates how large-scale digital identity infrastructure can support authentication, onboarding, and payment-linked services at national scale, making it relevant to future agentic commerce architectures.
• Singapore Singpass: 5 million active users provide another template for digital ID integration with payment systems;
• IMDA (Singapore): Published a framework requiring risk assessments and human oversight checkpoints for autonomous agent deployments.

The Attribution Problem is Underrated Aspect of Agentic Commerce

One of the most consequential and simultaneously underreported challenges of agentic commerce is attribution. In traditional e-commerce, retailers track the full user journey: impressions, clicks, browse behavior, add-to-cart events, and conversions. In agent-mediated commerce, the discovery, browsing, comparison, and preference-formation phases all occur inside ChatGPT, Gemini, or Perplexity. The merchant’s data stream begins only at the add-to-cart moment.

This creates a structural tension: merchants gain access to AI-driven discovery at unprecedented scale, but they know practically nothing about their actual customers. They lose the behavioral data that powers personalization, retargeting, and media measurement.

At the same time, if the mentioned digital ID systems are combined with KYA frameworks, agentic transactions happening on behalf of the individual may be somehow linked to the user behind the intent, creating new marketing databases. However, whether those might be eventually used for personalized offers and targeted ads, sellers are used to, remains a  rhetorical question. In any case, solving the attribution gap is one of the defining technical challenges for the payments and commerce industry in the next 18 months.

Conclusion

Agentic payments in e-commerce are currently turning into a massive market potentially growing at over 200% yearly rate. The infrastructure window, i.e. the period in which early-moving PSPs, merchants, and payment platforms can establish positions in the agentic payment stack, is open right now, in 2026.

However, the trust gap signals that full autonomy in online shopping and B2B money transfers is not yet the mass-market reality. At the same time, the future trajectory of agentic payments is clear: agents will begin with low-stakes, repetitive purchases and expand as standards mature and trust is established. By 2028, the platforms that built agent-compatible infrastructure in 2026 will be processing a disproportionate share of digital commerce.

The companies that will lead in agentic payments are those that invest now in agentic payments protocols integration, tokenized agent identity, fraud models calibrated for non-human transaction patterns, agentic payment orchestration systems, and the data infrastructure to survive the attribution gap.