Visa’s entire business model is betting on human behavior. It’s about human consumption and psychology. The reward points you accumulate, the fraud protection you rely on, the Centurion card you dream of, and the zero-liability policy that gives you peace of mind when swiping your card at foreign ATMs—all of this exists not because money transfer is difficult, but because humans are anxious, seek status, and are not good at reading terms and conditions. Visa has leveraged this cognitive difference to build a $500.00B company.

However, AI agents do not possess these traits. They don’t accumulate points, don’t seek fraud protection, and don’t aspire to own a black card. They have only one instruction: complete the task. And when the task involves payment, the agent performs complex calculations that humans would never bother to calculate: the cheapest path, the fastest settlement, the lowest fees. Every time, automated, without any emotion.

Last month, an article on SubStack titled “Global Intelligence Crisis in 2028” caused Visa’s stock price to plummet 4.00%, Mastercard to fall 6.00%, and American Express to fall 12.00% in a single day of trading. The report was labeled as a “scenario analysis” rather than a “prediction.” But the market didn’t buy it. The technical arguments were also irrelevant. The problem is that by 2027, agents will bypass transaction centers and instead use stablecoins for settlement. Visa has spent fifty years perfecting its product, and now its customer base is being replaced. In machine-to-machine commerce, a 2-3% interchange fee is clearly a target. This is the core argument of Citrini Research.

This does not mean that artificial intelligence will destroy Visa tomorrow. Rather, the fee structure on which Visa has built its business empire is essentially a tax on human irrationality, while the traders themselves are completely rational. This is what Visa is all about.

What is Visa selling? To understand why this is important, you must understand what the interchange fee is actually used for. When you use a credit card to shop, the merchant pays a 2-3% fee to the credit card network and your issuing bank. This fee is used to pay for your points rewards, fraud protection, shopping insurance, and dispute resolution services. The entire consumer value proposition of credit cards is borne by the merchant, who ultimately passes the cost on to consumers by slightly increasing the price of goods. This is a perfect and stable system that has been running for fifty years, because consumers in the transaction are willing to bear all these costs, except that they do not pay directly.

AI agents don’t need these. It will not object to the fee, nor will it request a refund. The rationale for charging this fee is that it protects against human error, fraud, and impulsive behavior. If there are no humans involved in the transaction, this fee is completely meaningless.

American Express is the most typical example of this problem. Its customers are high-income, high-spending, and aspirational high-end cardholders. Its annual fee is higher than Visa or Mastercard, precisely because its customers are willing to pay for identity and privilege. The premise of this model is that purchasing behavior is human-driven, and customers choose American Express over Visa because access to VIP lounges is worth the money. Agents will not actively choose American Express; they will only look for the cheapest solution to complete the transaction. In a world where software controls credit cards, high-end membership levels do not exist.

An agent-led commercial routing model that bypasses interchange fees poses a greater risk to credit card banks and single-business issuers that rely heavily on 2-3% transaction fee revenue and build their entire business segment around merchant-subsidized reward programs. Visa and Mastercard have network businesses that can adapt. But those issuers who build their entire profit and loss model around interchange fees and reward programs have nowhere to retreat.

The week everyone shipped at once: The Citrini report and the infrastructure project launch happened to be released within the same three weeks. Tempo officially launched on the mainnet last Wednesday. The payment blockchain jointly developed by Stripe and Paradigm, designed for high-volume stablecoin settlement, was launched in sync with the Machine Payment Protocol (MPP). MPP is an open standard that allows AI agents to autonomously pay for services without manual approval. The protocol introduces a session mechanism. The agent only needs to authorize a spending limit once, and then it can continue to make micro-payments when consuming services such as data, computing, or API calls. Funds are paid using OAuth authentication. The user authorizes a budget, and the agent can spend it. The entire process does not require using a bank card at every step.

Anthropic, DoorDash, Mastercard, Nubank, OpenAI, Ramp, Revolut, Shopify, Standard Chartered Bank, and Visa are all listed as design partners for Tempo. The entire payment and e-commerce ecosystem recognizes this structural change. On the same day that Tempo went live, Visa’s cryptocurrency division launched a command-line interface tool for AI agents to make payments through terminals without API keys, accounts, or manual authorization. Visa calls it “command-line commerce”—machines can transact without human intervention. Mastercard agreed to acquire stablecoin infrastructure startup BVNK for $1.80B. Circle launched Nanopayments on the testnet, a sub-cent and gas-free USDC transaction designed for agents to use pay-per-use APIs without accounts or credentials. Sam Altman’s World project launched AgentKit, which allows agents to carry cryptographic proofs to prove that they represent real people. The toolkit is directly integrated into Coinbase’s payment system, enabling the platform to verify agent identities without hindering legitimate transactions.

In my opinion, what happened last week was that companies were racing to become the new Visa, lest Visa realize what it had lost.

The obvious paradox: Now nothing is said clearly enough, that is, Visa is not stagnant. It participated in the development of Tempo’s Machine Payment Protocol (MAPPS), launched Visa Crypto Lab, and its cryptocurrency head also wrote an article in Fortune explaining how agents can use bank cards to pay through new standards. Mastercard is investing $1.80B in stablecoin infrastructure. Stripe acquired Bridge and Privy. Existing companies are aware of this shift and are preparing for the full arrival of the new infrastructure. Visa’s argument is that it can extend its track to agent-driven commerce before agent-driven commerce builds a track that makes Visa irrelevant. This statement is not entirely wrong. Stripe processed $1.90T in total payments in 2025, a year-on-year increase of 34.00%. These companies are not shrinking. The network distribution advantages of card organizations are difficult to replicate.

I admit that I am not very willing to say this publicly, because historically, once someone puts forward this argument, a new product will be released to make them look stupid. So, the loophole in this argument is here: Visa’s distribution advantage is based on its relationship with merchants and consumer trust. Merchants accept Visa because consumers hold Visa; consumers hold Visa because merchants accept Visa. The operation of the entire cycle depends on people. Once agents become the main buyers in an important business field, this flywheel will slow down. Agents have neither brand loyalty nor wallets. All they have are budgets and instructions. Whichever route is the cheapest and fastest will win their business, and the switching cost is zero.

I want to accurately state where we are, because the current development speed of public opinion has exceeded the data itself. Although the ecosystem around x402 is valued at approximately $7.00B, on-chain data shows that the protocol’s daily transaction volume last week was only about $28000.00, most of which came from testing rather than actual transactions. This number is vastly different from Visa’s daily transaction volume. The transaction volume of x402 has exceeded 50.00M. Although the amount of a single transaction is small, the number of transactions indicates that the infrastructure is being used. Developers are developing based on this. Merchant-side services that accept agent payments are also growing. This is how payment networks start.

McKinsey estimates that by 2030, AI agents may facilitate $3.00T to $5.00T in global consumer transactions. This estimate may be correct or too optimistic. But what is undeniable is that the agent-driven business model has not yet been widely adopted. Merchants building native agent services, companies using agents as primary buyers, and transaction volumes that can truly test the economics of transactions are still developing. The reason why Citrini’s report caused market panic is that it simulated a series of credible events. Mastercard’s Q1 2027 financial report will not attribute the slowdown in transaction volume to “agent-led price optimization.” At least not for now. The first to be affected will be micro-payments for AI infrastructure, not consumer commerce. Agents completing research tasks will call hundreds of specialized data APIs per session. Each call costs only a fraction of a cent. After a week, these calls may bring $40.00 in revenue to the developers operating the service. Credit card networks cannot cope with this situation. The economics of minimum transaction amounts do not work. The merchant onboarding process does not work. The fee structure does not work. This type of business model is destined to be unable to operate within Visa’s framework. It needs a completely new model, and x402, Nanopayments, and Tempo are building this model.

As the model built by Citrini shows, the disruption to consumer commerce, even if it happens, will only come later. It requires agents to handle a considerable portion of discretionary spending, which in turn requires consumers to trust agents and hand over the purchasing decisions they currently make themselves to agents. Visa has been hit by higher-quality customers. These customers no longer need the elements that Visa relies on to succeed. The 2-3% interchange fee is not a transaction tax, but a tax on human irrationality. And agents are completely rational. How do I know this is important? Because Visa spent $1.80B last week to ensure that it is not excluded from the answer.

[Block unicorn]