Visa and Mastercard Enter Agentic Commerce War

Two weeks ago, Mastercard expanded its Agent Suite with new merchant-facing capabilities via Merchant Cloud that allow AI agents to initiate, authorize, and complete transactions on behalf of consumers.

On June 10, 2026, Visa, its biggest competitor, announced a strategic collaboration with OpenAI. This partnership integrates Visa's global payment network into OpenAI experiences. It covers tokenization, agent identification, real-time authorization, and fraud monitoring for AI-initiated transactions at scale.

The pattern across both announcements is the same. The world's two largest payment networks are now building simultaneously for agentic commerce, and the race to become the default trust layer between AI agents and the financial system is underway.

"AI will transform commerce more profoundly than the internet or mobile technology ever did," said Jack Forestell, Chief Product and Strategy Officer at Visa. "As AI agents become active participants in the economy, Visa's focus is to ensure transactions are trusted, secure, and seamless. That's the infrastructure we're building with partners like OpenAI."

The Infrastructure Race

The Visa-OpenAI announcement is the higher-profile of the two moves. OpenAI's platforms are used by millions of people daily, and ChatGPT's expanding suite of agentic capabilities represents the consumer-facing interface where AI commerce is most visible to mainstream users.

But the significance of the announcement extends beyond ChatGPT shopping. It is about which payment network gets embedded at the foundational layer of the agentic commerce ecosystem before that ecosystem matures.

Visa's global network processes more than 300 billion transactions annually. The same tokenization, authorization, and fraud monitoring capabilities that underpin that volume are now being extended natively into AI-powered commerce environments.

For the global payments leader, the partnership is not a new product but an extension of existing infrastructure into a new interface layer. For OpenAI, access to global payment rails makes AI-initiated commerce credible to consumers and merchants, who need to trust that an AI agent executes transactions with the same level of security as a human checkout, according to the press release.

Mastercard's approach through Agent Suite is structurally similar. It embeds existing payment infrastructure into AI-driven merchant workflows rather than building a new payment system for AI.

The competitive dynamic between the two networks in agentic commerce is now visible for the first time. Mastercard moved first with its merchant-focused Agent Suite. Visa responded by announcing a platform-level partnership with the world's most widely used AI platform.

Both are moving to be embedded in the agentic layer before any single platform or protocol becomes dominant.

The timing compression is notable. Mastercard expanded its Agent Suite with new merchant-facing capabilities on May 26, 2026. Visa announced the partnership with OpenAI on June 10, 2026.

About 15 days ago, two major moves in payment infrastructure highlighted how urgently both networks are acting to establish a position in the still-evolving commerce layer.

The Consumer Control Architecture

The most technically specific part of the Visa-OpenAI announcement is the permission framework governing what AI agents can and cannot do with a consumer's payment credentials.

Visa's approach uses tokenized credentials, replacing sensitive card data with secure network tokens bound to specific agents and specific use cases rather than to a card number.

A token issued for a grocery shopping agent cannot be used for travel bookings. A token with a $200 spending limit cannot authorize a $500 transaction. These are technical constraints enforced at the network layer, applied in real time during authorization.

Transactions operate within user-defined permission controls that cover spending limits, merchant category restrictions, and required approval thresholds.

A consumer can instruct their AI agent to execute routine grocery reorders autonomously while requiring explicit approval for any single purchase above a set threshold. The agent does not bypass that threshold. The Visa infrastructure enforces it, according to the press release.

"By integrating with Visa Intelligent Commerce, we're building the infrastructure for secure, transparent, and user-controlled agentic transactions, helping people do more with AI agents while maintaining confidence that payments are being handled safely and securely," said Marco Mahrus, Head of Partnerships, Commerce at OpenAI.

The consumer remains in command of the parameters. The agent operates within them. The payment network enforces the boundary between the two.

Mastercard's Agent Suite addresses the same problem through a similar architecture, on-behalf tokenized credentials, consent-based payments, and authentication controls that keep merchants and consumers in control of what agents can do, according to the company. The two networks are independently converging on the same governance model.

The Enterprise and B2B Dimension

Consumer shopping is the visible application in the Visa-OpenAI announcement. The more commercially significant long-term dimension is the enterprise and B2B layer.

The partnership explicitly extends into developer-focused applications powered by Codex, OpenAI's coding agent, covering procurement, invoicing, reconciliation, and payment execution in automated and conversational business workflows, according to the press release.

These are the financial operations that enterprises run at scale, where AI agents taking autonomous action on behalf of procurement teams, finance departments, and operations managers represent a fundamentally different efficiency opportunity than a consumer asking ChatGPT to reorder household goods.

The Codex integration signals that Visa is positioning its infrastructure within the enterprise AI workflows being built now, rather than just the consumer commerce experiences that are further from mainstream adoption.

For enterprise AI platforms building automated financial workflows, Visa's tokenized payment infrastructure and real-time authorization capabilities are a required component.

OpenAI filed confidentially for an initial public offering the same week as Visa's announcement, with rival Anthropic filing separately during the same period, according to WSJ.

The timing places the Visa partnership within OpenAI's pre-IPO narrative, establishing that OpenAI has the payment-infrastructure relationships needed to make its agentic commerce ambitions credible to public-market investors.

A commerce platform without payment rails is a discovery tool. A commerce platform with Visa's global network embedded inside it is a transaction platform. The distinction matters as OpenAI prepares to tell public market investors a story about its revenue model beyond subscriptions and API access.

The key question is no longer whether AI agents will participate in commerce. Visa and Mastercard are already building for that future. The question is who controls the trust layer that governs how agents spend money, which rules they follow, and which network becomes the default infrastructure for an economy in which software increasingly acts on behalf of humans.