Artificial intelligence agents are about to take market share away from Visa

Byline: Thejaswini M A

Compiled by: Block unicorn

Introduction

Visa’s entire business model is built on betting on human behavior. It’s about human consumption and psychology. The rewards points you rack up, the anti-fraud protection you rely on, your coveted Centurion card, and the zero-liability policy that makes you feel secure when swiping your card at ATMs abroad—none of this exists because moving funds is difficult. It exists because people are anxious, chasing status, and not good at reading the terms and conditions. Visa has leveraged this cognitive gap to build a company with a market value of $500 billion.

However, AI agents don’t have these traits.

They don’t accumulate points, they don’t pursue anti-fraud protection, and they don’t long for black cards. They have one instruction: complete the task. And when the task involves payments, the agent will perform complex calculations that humans will never bother to compute: the cheapest route, the fastest settlement, and the lowest fees. Every time—automatically, with no emotion.

Last month, a SubStack article titled “The Global Intelligence Crisis of 2028” caused Visa’s stock price to plunge 4% in a single trading day, Mastercard to fall 6%, and American Express to drop 12%. The report was labeled as a “scenario analysis,” not a “forecast” (as stated in the original). But the market isn’t buying it. The technical claims don’t matter either. The real issue is that by 2027, agents will bypass the transaction centers and switch to settling with stablecoins. Visa spent fifty years perfecting its product, but its customer base is now being replaced.

In machine-to-machine commerce, a 2–3% interchange rate is obviously a target. That conclusion from Citrini Research is at the heart of its core argument. This doesn’t mean AI will destroy Visa tomorrow. Instead, the fee structure that underpins Visa’s business empire is, in essence, a tax on human irrational behavior, while traders themselves are perfectly rational. That’s what Visa’s existence is really about.

What is Visa selling?

To understand why this matters, you need to know what interchange fees are actually used for.

When you shop with a credit card, the merchant pays a 2–3% fee to the credit card network and to your issuing bank. This fee goes toward paying for your rewards, anti-fraud protection, purchase insurance, and dispute resolution services. The entire consumer value proposition of credit cards is borne by the merchant, and the merchant ultimately passes the costs on to consumers by slightly increasing the price of goods. It’s a refined and stable system that’s been running for fifty years, because consumers in transactions are willing to bear all those costs—even if they don’t pay them directly.

AI agents don’t need any of this. They won’t object to the fees, and they won’t request refunds. The rationale for charging this fee is that it can guard against human error, fraud, and impulsive behavior. If no human is involved in the transaction, then the fee is completely meaningless.

American Express is the clearest example of this problem. Its customers are high-income, high-spending, and ambitious premium cardholders. Its annual fee is higher than Visa’s or Mastercard’s—because its customers are willing to pay for identity and privileges. The premise of this model is that purchasing behavior is human-driven: customers choose American Express over Visa because access to VIP lounges is worth it. Agents won’t proactively choose American Express—they’ll simply look for the cheapest option to get the transaction done. In a software-controlled credit card world, premium membership tiers don’t exist.

The business routing model led by agents that bypasses interchange fees creates a bigger risk for credit card banks and single-product issuers that rely heavily on 2–3% fee revenue and build their entire business segment around reward programs funded through merchant subsidies. Visa and Mastercard have network businesses that can adapt. But those issuers that built their entire profit-and-loss model around interchange fees and rewards programs have nowhere to go.

The week when everyone ships at the same time

Citrini’s report and the infrastructure projects were released within the same three-week window.

Tempo officially launched its mainnet last Wednesday. The payment blockchain co-developed by Stripe and Paradigm—built specifically for high-volume stablecoin settlement—launched in sync with the Machine Payment Protocol (MPP). MPP is an open standard that allows AI agents to pay service fees autonomously, without needing to have a human approve each one individually. The protocol introduces a session mechanism. An agent only needs to authorize a single spending cap, and it can keep making micro-payments as it consumes services such as data consumption, computation, or API calls. Funds payments use OAuth authentication. The user authorizes a budget, and then the agent can spend it. The entire process doesn’t require using a credit card at every step.

Anthropic, DoorDash, Mastercard, Nubank, OpenAI, Ramp, Revolut, Shopify, Standard Chartered, and Visa are all listed as Tempo design partners. The whole payments and e-commerce ecosystem recognizes this kind of structural change.

On the very same day Tempo went live, Visa’s crypto division released a command-line interface tool that enables AI agents to make payments through terminals, with no API keys, no accounts, and no human authorization. Visa calls it “command-line commerce”—transactions can happen without any human intervention by the machines.

Mastercard agreed to acquire the stablecoin infrastructure startup BVNK for $180 million. Circle rolled out Nanopayments on its testnet—an agent-built, per-use API that works without accounts or credentials, with transaction costs below a cent and no Gas fees, for USDC. Sam Altman’s World project launched AgentKit, enabling agents to carry cryptographic proofs to prove that they are acting on behalf of real humans; the toolkit is directly integrated into Coinbase’s payment system, allowing the platform to verify agent identities without blocking legitimate transactions.

In my view, what happened last week is that companies raced to become the new Visa, so Visa wouldn’t realize what it has already started to lose.

An obvious paradox

There’s nothing unclear about this: Visa hasn’t been standing still.

It has been involved in developing Tempo’s machine payment protocol (MAPPS), launched Visa Crypto Labs, and its crypto head also wrote in Fortune explaining how agents can pay with cards through the new standard. Mastercard is investing $180 million into stablecoin infrastructure. Stripe acquired Bridge and Privy. Existing companies have already recognized this shift and made preparations before the new infrastructure arrives in full.

Visa’s argument is that it can extend its rails into agent-driven commerce before something arises that makes Visa irrelevant within agent-driven commerce.

This claim isn’t entirely wrong. Stripe handled $1.9 trillion in payments in 2025, up 34% year over year. These companies aren’t shrinking. The network distribution advantage of card organizations is hard to replicate. I admit I’m not too keen on saying this publicly, because based on historical experience, once someone makes this argument, new products get released that make them look foolish.

So here’s the flaw in the argument: Visa’s distribution advantage is built on relationships with merchants and consumer trust. Merchants accept Visa because consumers hold Visa; consumers hold Visa because merchants accept Visa. The whole loop depends on humans. Once agents become the primary buyer in an important segment of commerce, this flywheel slows down. Agents have no brand loyalty and no wallet. All they have is budgets and instructions. No matter which route is cheapest or fastest, they can win their business, and switching costs are zero.

I want to state exactly where we are, because the pace of public opinion right now is already outstripping the data itself.

Even though the ecosystem valuation around x402 is roughly $7 billion, on-chain data shows that the protocol’s daily transactions last week were only about $28,000, most of which came from testing rather than real transactions. Compared with Visa’s daily transaction volume, this number is worlds apart.

x402’s transaction volume has already surpassed 50 million. Even though the transaction amount per transfer is small, the number of transactions shows that the infrastructure is being used. Developers are building on top of it. Services on the merchant side that accept agent payments are also steadily growing. That’s how payment networks get started.

McKinsey estimates that by 2030, AI agents could enable $3 trillion to $5 trillion in global consumer transactions. That estimate might be correct, or it might be overly optimistic. But there’s no denying that agent-driven business models haven’t yet been widely adopted at scale. Merchants building native agent services, companies using agents as their main buyers, and transaction volume that can truly validate the economic efficiency of transactions are still in development.

Citrini’s report triggers market panic because it simulates a series of plausible events. Mastercard’s Q1 2027 earnings won’t blame the slowdown in transaction volume on “agent-driven price optimization.” At least not for now.

First affected is not consumer commerce, but micro-payments in AI infrastructure.

Each agent that completes a research task makes hundreds of calls to specialized data APIs per session. The cost of each call is only a fraction of a cent. Over the course of a week, these calls could bring in about $40 in revenue for the developers operating the service. The credit card network can’t handle this. The economics of a minimum transaction amount model don’t work. Merchant onboarding processes don’t work. The fee structure doesn’t work. These kinds of business models are destined not to run within Visa’s framework. It requires an entirely new model, and x402, Nanopayments, and Tempo are building it.

As shown in the model Citrini constructed, even if the disruption of consumer commerce happens, it will come even later. It requires agents to process a significant portion of discretionary spending, which in turn requires consumers to trust agents—handing over the purchase decisions they currently make themselves.

Visa is being hit by higher-quality customers. These customers no longer need the elements that make Visa successful. A 2–3% interchange fee isn’t a transaction tax—it’s a charge on human irrational behavior. And agents are completely rational.

How do I know this is important? Because last week Visa spent $180 million to make sure it doesn’t get left out of the answer.