Visa's stablecoin and agentic commerce initiatives lack revenue clarity in Q3 FY2026 earnings.

Written by: Will A Wang

On July 16, Visa launched the Visa Stablecoin Platform, a platform for minting, transferring, and managing stablecoins, with its inaugural coin being OpenUSD—a stablecoin initiated by a consortium of 140 major institutions. That week, the news dominated payments and crypto circles: a network processing nearly half of the world’s card transactions announced it would provide the operational infrastructure for stablecoins.

On the Q3 FY2026 earnings call, 12 days from now, Visa CEO Ryan McInerney addressed global investors with a comprehensive overview of this platform—covering its features, positioning, initial supported tokens, and future integration with Pismo to support tokenized deposits. However, not a single word was mentioned regarding how it will be priced, which revenue line item it will fall under, when it will go live, or who the first customers will be. In CFO Christopher Suh’s financial segment, the letters VSP never appeared at all.

The more thoroughly you describe the product, the more noticeable the absence of money becomes.

The Q&A session made this even clearer: Of the 13 analysts, only Jeff Cantwell from Seaport asked about stablecoins—specifically whether OpenUSD was targeting Circle and Tether. The rest of the questions focused on cross-border growth rates, FIFA’s one-time boost, value-added services, customer incentives, and the annual recurring question about the revenue formula. The two analysts who gave Visa the highest price targets asked about the stage of internal AI adoption and promotional timing disruptions.

The people who market and feature VSP as a headline didn’t consider it an issue. And McInerney’s first response to Cantwell also cast a shadow over stablecoins. Instead of directly addressing competition, he said something else: stablecoins have not yet achieved real scale beyond a few use cases—such as the stablecoin-linked cards we’ve issued across the country.

This article does one thing: it takes every action Visa has taken related to stablecoins and agentic technology, and measures each one against its revenue statement. Once measured, you can see their strategic position within Visa and how much the narratives around these two terms have been amplified.

The CFO discussed only one thing at the earnings call: the revenue line—how much each line increased, what drove it, and which metric it was tied to. Stablecoins did not appear in his script for one reason: they don’t belong to any of those lines.

So, in this chapter, let’s first establish a benchmark: Who is Visa, how many fee points are there, and how much is taken from every 100 yuan?

If you think of Visa as a highway, everything else becomes easier to understand. It doesn’t produce cars, drive them, or own the warehouses at either end. It builds the road, sets the rules, and collects tolls at the exits. Each transaction on the road involves four parties: the cardholder, the issuing bank (the cardholder’s bank), the merchant, and the acquirer (the service provider on the merchant’s side). Visa is none of these four, but it is the road all four must travel together.

To understand how transaction fees are split, start with the total cost. The total cost a merchant pays for a transaction is called the Merchant Discount Rate (MDR), which in the U.S. is roughly 2% to 3%, divided into three parts: the largest portion is the interchange fee, going to the issuing bank; a portion is the acquirer’s markup; and only a small remaining share is the network fee, going to Visa. Although Visa sets the interchange fee, it keeps none of it: the fee flows from the acquirer to the issuer, compensating the issuing bank for bearing risks such as chargebacks, fraud, and funding costs—this is precisely why global regulators continually scrutinize it.

The merchant bears the cost, while cardholders are incentivized.

Card networks essentially bring together issuers and acquirers into a coalition of shared interests—standards merely provide the connection; it is the profit-sharing mechanism that unites them.

The entire system operates on the same logic: let others take the biggest share. That single “beep” on your phone means the wallet provider takes a cut from the issuing bank; Stripe and Adyen, leveraging their merchant relationships, run their own fleets along the route; and those fintech companies touting “disrupting payments” are all driving on Visa’s license plates.

Visa has turned every potential disruptor into a paid driver on its own path.

1.1 Four fee channels for buying a cup of coffee

A Hong Kong user bought a cup of coffee at a convenience store in Tokyo for $10, using a Visa card issued in Hong Kong.

Assuming the merchant discount rate for this store is 2.5%, the merchant receives $9.75 and pays $0.25. The majority of this $0.25 is the interchange fee, going to the Hong Kong-issued card issuer—which bears the user’s credit and fraud risk; a portion goes to the Japanese acquirer; and Visa takes approximately $0.09. (Visa does not disclose itemized rates; actual pricing varies by region, card type, and merchant category. The figures below are inferred from publicly available totals, assuming all components were explicitly stated, and can be recalculated by the reader.)

These nine cents were not collected all at once; they were collected separately at four different toll booths.

First channel: Service Revenue. This is not charged for the specific transaction, but for the fact that the card is accepted at all. The same card can be used at a convenience store in Tokyo, a website in New York, or a subway turnstile in Shenzhen—the existence of this acceptance network is the product itself, and fees are settled based on the issuing bank’s transaction volume. It is charged as a percentage of the transaction amount: the more expensive the coffee, the more this channel earns.

Second, data processing revenue. During those one or two seconds of swiping, the authorization request travels from the POS terminal to the acquiring bank, to VisaNet, to the issuing bank, and back along the same path; additionally, settlement and netting occur overnight.

This fee is a fixed amount per transaction. Annual data processing revenue is approximately $20 billion, handling about 29 billion transactions, or roughly 7 cents per transaction. Whether the coffee costs $10 or $1,000, the 7-cent fee remains the same.

Another source of growth unique to this channel is penetration: attracting vehicles already on the road onto our Visa network. Five years ago, Colombia’s processing penetration was still in the single digits; today, it exceeds 90%, and this growth requires no additional spending from consumers.

Third slot: International Transaction Revenue. This coffee was purchased in a country that uses yen using a Hong Kong dollar card, triggering the most expensive slot. It’s expensive because it includes both cross-border fees and currency conversion: Visa must process the transaction and convert between Hong Kong dollars and Japanese yen, with foreign exchange (FX) being the primary source of profit here. This slot is also charged as a percentage of the transaction amount, so of those 9 cents, this component contributes significantly more than it would for a same-amount local transaction.

Fourth slot: Other Revenue. If this transaction has been reviewed by risk management, if the card issuer’s account is processed on Pismo, and if the card’s marketing campaign was run by Visa: this slot charges based on services rendered, not on transaction amount or volume, and has nothing to do with the coffee itself. Risk and security (Featurespace), card processing (Pismo, DPS), acquiring solutions, consulting, and marketing services all fall into this category.

Also, clarify a commonly confused metric: Value-Added Services (VAS) is a business metric spanning four categories (risk control is included under data processing, while consulting and marketing are included under other income), whereas “other income” is merely one line item on the financial statement—the two are not equivalent. Please view both figures together under section 1.3.

Looking back at the 9 cents: the 7 cents from the second fee is fixed, with approximately 2 cents coming from the percentage-based fees. For comparison, Visa’s average transaction is $56, and at its overall net rate of 0.29%, it earns about 16 cents—also 7 cents fixed and 9 cents percentage-based.

The same fee structure charges 0.29% on a $56 transaction, but effectively 0.9% on a $10 coffee—smaller transactions are impacted more by the $0.07 fixed fee.

1.2 Token Following the Vehicle

There’s another thing that travels with your vehicle without occupying a toll lane: tokenization. This system, called the Visa Token Service, replaces the 16-digit card number on your card with a unique substitute number that is valid only in specific contexts.

Take this user, for example. He linked his card to Apple Pay, and when he did, Visa issued a token valid only for that device. He also saved his card on an e-commerce website, where another token was generated. Three months later, when the card expired and was replaced with a new one, the physical card number changed, but both tokens were automatically updated—requiring no re-linking from him. And if that e-commerce site were breached, the leaked data would be a string of numbers useless anywhere else.

Each issuance, update, verification, and routing represents an additional toll action on the road, with charges applied to the second gateway. Approximately half of global e-commerce transactions are now completed via tokens.

The credentials issued by Visa to AI agents also go through this same infrastructure.

1.3 Customer Incentives and What This Quarter's Metrics Revealed

After collecting the four fees, subtract one more item: Client Incentives—the amount refunded to the card-issuing bank and the acquirer.

The rebates negotiated during renewals with major issuing banks and the signing terms for new customers are all included here. Crucially, incentives are deducted from revenue, not recorded as costs, so Visa reports "net revenue." This quarter, incentive growth increased by 18%, a 4-percentage-point jump from the previous quarter.

After collecting the four fees and deducting the incentives, divide the net income by the payment amount:

0.29%, or 29 basis points. For every $100 transaction, Visa ends up with 29 cents. We will refer to it as 29 basis points throughout the rest of the text.

The previous full fiscal year also aligns: net revenue of approximately $40 billion, annual payment volume of $14.2 trillion, and 28 basis points.

The key point about this number is that it hasn’t changed in ten years. Over the long term, Visa’s gross toll rate has remained stable at around 40 basis points of transaction volume; during the same period, the portion returned as incentives rose from approximately 17% to about 29%, nearly doubling. With these two forces squeezing from both sides, the net toll rate has barely budged. In other words, the additional revenue generated over the years from cross-border and value-added services—those higher-priced lanes—has not become extra profit for Visa; instead, it has been passed on as rebates to traffic drivers.

This is the key to understanding Visa: it’s not raising toll fees, but rather using changes in traffic flow structure to offset rising acquisition costs, keeping the overall rate stable.

A company that has survived for decades by maintaining balance will not alter that balance for something new.

Visa has set the standard: one path, four tolls, twenty-nine basis points.

To date, Visa’s significant interactions with stablecoins have been limited to two areas: consumer-facing U cards and institutional settlement—how institutions settle accounts between themselves. These two operate differently. More importantly, they exist in entirely separate domains.

2.1 The essence of a U card is still a card transaction.

What happens at the moment you swipe a Visa card linked to a stablecoin balance—commonly known in the industry as a U card?

The network reads the BIN and routes the authorization request to the corresponding issuer. The issuer queries the underlying crypto platform: Does the user have sufficient funds? If so, the platform immediately calculates how much stablecoin needs to be sold to cover the purchase and completes the conversion within the authorization window. Once the fiat amount is confirmed, the issuer approves the transaction, and the merchant receives authorization.

The entire process is completed in milliseconds. For merchants, it’s indistinguishable from any ordinary Visa transaction—they receive fiat currency and need to make no changes to their payment processing code.

This is the most critical feature of the U Card: the on-chain portion occurs entirely before Visa's involvement.

A U Card cannot be accomplished by a single company alone. It requires at least several key components: card network membership and BIN, on-chain custody, real-time liquidity for conversion, and a full suite of KYC/AML and compliance capabilities. The traditional approach is to partner with a licensed institution that acts as a BIN sponsor, providing membership and card number ranges, while the crypto project white-labels the card on top. This path avoids the years-long process of applying for licenses independently.

In the past two years, a more advanced model has emerged: full-stack card issuers like Rain and Reap have become Visa Principal Members themselves—bypassing bank intermediaries entirely and acting as the issuer of record, settling directly with Visa. Rain alone supports over 200 card programs.

How is the money on this chain distributed?

The issuer receives the interchange fee, typically 1% to 2% of the transaction amount, which is either returned to users as cashback or retained as profit. The conversion party earns revenue from FX spreads and exchange rate differentials. The project party may also charge cardholders annual fees and withdrawal fees. Visa receives the additional 29 basis points beyond these fees.

So where did the U card land? Go through all four toll booths, because it was just a regular card transaction. After the conversion is complete, what hits the road is fiat currency—the toll booths have no idea the stablecoin ever existed.

Visa has documented two settlement models for its U card: In the traditional model, the project first converts stablecoins into fiat currency before settling with Visa, using the blockchain solely as a ledger; in the emerging model, the Principal Member settles directly with Visa in USDC, after which Visa’s digital custodian converts the funds into fiat and pays the acquirer. In either case, only the latter portion—the way institutions settle their accounts—changes.

This is the most significant counter-evidence: the only revenue-generating stablecoin activity is the conversion of stablecoins back into card transactions.

2.2 The Backstage Path: Settlement

Visa cards handle consumer payments, while stablecoins manage fund transfers—each layer of infrastructure takes care of its respective segment.

The two paths do not intersect. Stablecoins appear on these paths in different ways: on the front-end spending path, they must first be converted back into fiat currency to enter the VisaNet network, and all four toll booths mentioned in Chapter One are located on this path; on the back-end settlement path, they can directly serve as clearing assets to settle the net difference between issuers and acquirers.

A card transaction feels like a one- or two-second authorization to the consumer, but that’s just the flow of information. The actual money moves between the issuer and the acquirer, with Visa’s treasury and settlement system standing in between. Each day, it nets all receivables and payables for each member institution into a single net amount, which is then settled in one transaction. This net amount is what Visa officially refers to as VisaNet obligations.

Traditionally, this net amount settles through fiat systems—ACH, Fedwire, SEPA, or local clearing—which operate only during banking hours. What stablecoin settlement changes is just one thing: the asset used to settle this net amount, nearly instantly and around the clock.

Someone has to step in because of the time lag—and that costs money. Transactions authorized on Friday night won’t settle until at least Monday, so someone must cover the funds in the interim. But not everyone can cover these funds—Visa’s rules require customers with insufficient credit to provide collateral. Those with strong credit typically don’t need to post collateral, but crypto issuers with high card volumes but average credit must hold their funds in reserve and cannot use them for business operations.

The reason collateral exists is that Visa has promised its customers that it will cover any settlement obligations that a participant fails to meet. In FY2025, its average daily settlement exposure amounted to $91.2 billion. On the back-end, Visa doesn’t charge a toll, but it bears the guarantee.

2.3 From Stablecoin Settlement to VSP

Visa began its stablecoin pilot on the settlement side in March 2021. The first transaction was with Crypto.com, addressing a specific pain point: instead of selling crypto assets for USD and then wiring funds to cover Visa’s card program liabilities, switching to USDC allowed direct transfers to Visa’s custodial wallet, eliminating one conversion step and a wire transfer cutoff deadline. The pilot later expanded from the issuing side to the acquiring side (Worldpay, Nuvei), launched domestically in the U.S. in December 2025, and by April 2026 supported nine blockchains.

Visa’s list of benefits for this initiative is straightforward: funds available in seven days, uninterrupted processing over weekends and holidays, reduced collateral requirements due to seven-day settlement, automated treasury management, and simplified reconciliation. Most importantly, there’s the qualifying addendum—consumers’ card experience remains unchanged.

What about the scale? $3.5 billion annualized in November 2025, $4.5 billion in January 2026, and $7 billion in April—doubling in five months, that’s the growth rate.

Sheffield's denominator in his January Reuters interview was Visa's payment volume of $14.2 trillion in the prior year; dividing $7 billion by that figure yields 0.05%.

Convert assets used to settle VisaNet obligations from fiat to stablecoin, with Visa’s fees remaining unchanged. The benefits go to the issuing banks—shorter exposure windows, reduced collateral requirements, and the ability to settle over weekends. Merchant discount rates remain unaffected by faster settlement, and consumers notice no difference at checkout.

This is an operational improvement, not a revenue recognition. And what VSP aims to commercialize is precisely this journey.

The settlement pilot is for Visa’s own use, while VSP sells this capability to others: a new toll booth on the backend highway. Ideally, wallet-as-a-service should follow a service revenue model, while minting and transfers should follow a data processing model. However, Visa has not yet provided any billing methodology, accounting classification, or commercial timeline. The platform is in beta, and customers remain unnamed.

It's not that we can't accept it; it's just that we haven't started accepting it yet.

Overall, Visa earned only two payments today from stablecoins: one from converting stablecoins back into card transactions, and another from selling advice about stablecoins.

This strategy isn’t new at all. Looking back at Chapter One: Stablecoins, it’s just the latest name on the list of paid drivers. And to this day, not a single name on that list has opened a new fee channel for Visa.

Put the three actions back on two paths: U-card on the front-end path, and settlement and VSP on the back-end path. Since toll collection occurs only on the front-end, this is the sole reason why U-card generates revenue while settlement and VSP do not.

The stablecoin has been exhausted; the front-end portion is actually simpler—it all falls along the front-end path.

3.1 Frontend: The Device That Enables Agents to Get Started

Nothing new on the frontend: Every agentic product Visa creates is a device to get agents onto the old path.

The segmentation was defined by Visa itself. When VSP was launched in July, Chief Product and Strategy Officer Jack Forestell said AI is transforming the front end of commerce and stablecoins are reshaping the back end; three months later, at the earnings call, McInerney repeated this exact phrasing. This is not PR jargon—it’s an internal strategic division.

The product list is short: Agent Score, Agent Directory, Token Assurance Framework, plus partnerships with OpenAI and Meta. All of them operate through the fourth channel—billing details are undisclosed, but they sell capabilities such as scoring, directories, and credential management, charged per service rather than transaction value; they also rely entirely on the token infrastructure introduced in Chapter One, extended from mobile devices to agents.

3.2 Alternatives and Creation: Visa Only Saw Half

There is only one thing that cannot be handled by either the frontend or backend: transactions between agents.

Bernstein’s analysts asked this in person—how will agentic systems expand the addressable market, and will agent-to-agent interactions create new economic structures?

McInerney answered the first half of the question using the analogy of e-commerce, mobile commerce, tokenization, and contactless payments—each of which went through “establish standards first, then early adoption, followed by consumer momentum, and finally scaling.” His conclusion was that agentic commerce is a question of “when,” not “if.” He did not answer the second half of the question.

But that very analogy already answers it: e-commerce, mobile payments, tokenization, and tap-to-pay—all four precedents changed how people pay, not who is buying.

Agentic commerce, in fact, consists of two distinct things—a distinction we have repeatedly emphasized in our research on agentic payments:

One type is agency-based procurement. The IT agent automatically adds or removes SaaS seats based on employee onboarding and offboarding, while the travel agent books flights and hotels according to budget and policy. These transactions involve higher amounts, and the merchants and channels resemble those in traditional e-commerce—they are substitutive, capturing market share from existing human purchasing behaviors.

Another category is machine-to-machine procurement. A research agent pays a few cents to multiple data providers to retrieve financial reports for cross-verification; the main agent outsources subtasks to a retrieval agent, which then purchases reports from a data agent. These transactions are high-frequency, low-cost, and involve no accounts or contracts between buyer and seller—they create new relationships that simply did not exist before.

Visa’s analogy sequence covered only the first category from start to finish. Sellers used the same metric: Morgan Stanley’s forecast of $190 billion to $385 billion by 2030, based on the assumption that AI will account for 10% to 20% of U.S. e-commerce—measuring only AI replacing human online shopping, thus covering only the first half.

Visa isn't paving a new path for agentic; it's replacing the drivers on the existing road.

(Agentic Payment from Visa’s Perspective)

4.1 Tier 4, ranked equally with brand advertising

First, look at how high it prioritizes this matter. The answer is fourth tier, tied with brand advertising.

This quarter, Visa laid off approximately 2,600 employees, or about 7% of its workforce, primarily in technology and product roles, and recorded severance expenses of $563 million. KBW’s Sanjay Sakhrani asked whether the savings would flow into profits or be reinvested.

McInerney responded by reinvesting everything and then listed the investment directions from the beginning: consumer payments (expanding acceptance in cash-dominated markets, premium customer segments, cross-border e-commerce), value-added services (risk management and security, marketing services, Pismo, Featurespace), and business and fund flows (unified B2B, embedded finance, cross-border remittances), followed by “on top of that, stablecoins and agentic.”

The order of the top tiers varies slightly across reports, but stablecoins and agentic always appear at the bottom. This ranking, given by the CEO off-the-cuff when questioned about capital allocation, is more credible than any keynote speech.

Two additional confirmations are equally clear. First, they are not part of the growth engine—Visa’s 10-K and all past earnings calls consistently outline only three strategic pillars: consumer payments, commerce and funds flow, and value-added services. Stablecoins and agentic solutions are components embedded within these three pillars, not standalone elements; thus, the strategic framework provides no dedicated revenue channel for them.

Second, they are the only directions without an addressable market. Visa is a company heavily reliant on the TAM narrative: consumer payments, with annual addressable consumer spending exceeding $40 trillion (excluding China and Russia), of which the underserved portion exceeds $20 trillion; commercial and funds flows, with an annual payment opportunity of $200 trillion; and value-added services, with an annual potential revenue opportunity of $520 billion. Each segment has a trillion- or hundred-billion-dollar denominator, while stablecoins and agentic have no such figures.

4.2 Triple Jump: Settlement, VSP, and OUSD are on the same line

Tier four does not mean having no strategy. On the contrary, Visa's actions behind the scenes follow a strict sequence.

Level 1: Settlement pilot. The five-year-old line from Chapter Two. It doesn’t generate profit, but it has built an entire dedicated logistics infrastructure: custody, multi-chain asset transfers, fiat and stablecoin on/off ramps, and reconciliation.

Level 2, VSP. Open this dedicated lane for通行 and sell it to customers. Who are these customers? Chapter Two has already explained—they are companies that bulk-purchase cryptocurrencies but cannot enable users to spend them. Visa’s Global Growth Head, Rubail Birwadker, puts it even more clearly:

It’s less about acquiring stablecoins and more about how this integrates with clients’ treasury settlement, cash flow workflows, and existing banking setups.

This is an admission of a treasury business, not a declaration of a crypto business.

Level 3, OUSD.

Two details about OUSD: First, Visa secured a seat on the distribution channel, not ownership of the alliance—which is managed by Stripe’s team, with its most direct competitor sitting at the same table. Second, the initial cryptocurrencies supported by VSP are precisely OpenUSD.

The loop has closed: Visa paved its own settlement road, which opened as the VSP platform; and the first vehicles assigned to this road carry its own governance-issued license plate and stablecoin.

4.3 The neighbor spent 1.8 billion on a bridge

For the same question, Mastercard submitted a completely different answer.

Six days after Visa’s earnings call on August 3, Mastercard completed its acquisition of BVNK. Founded in 2021, BVNK processes approximately $30 billion in stablecoin payments annually across 200 countries and regions. The transaction value reached up to $1.8 billion—$1.5 billion in base consideration plus an additional $300 million in performance-based contingent payments. Although the deal was initially announced in March with an expected closing by year-end, it closed nearly five months ahead of schedule. One notable detail: among BVNK’s clients is Visa Direct—meaning the bridge Mastercard acquired was already carrying Visa’s traffic.

Put two numbers together: The volume of stablecoins purchased through Mastercard in a single day is more than four times what Visa generated over five years.

They are also different in nature: BVNK’s $30 billion represents customer payment volume that generates revenue, while Visa’s $70 billion refers to internal settlement flows that do not generate revenue.

Jorn Lambert, Chief Product Officer at Mastercard, stated: Stablecoins are addressing real needs in cross-border B2B, remittances, payroll, settlement, and treasury management. One is listing specific use cases; the other is emphasizing that scaling has not yet been achieved—within the same time frame and with the same industry standing.

The old platform emphasized compatibility; the new platform emphasizes alignment.

One company spent its money on its balance sheet, while the other spent it on its product roadmap.

(Mastercard acquires BVNK for $1.8 billion—not for the technology, but for time)

At this point, the strategy is clear: no token issuance, no liability, no custody of funds—build your own private lane on the backend path, and construct narrow ramps at every possible junction that might bypass you.

The question is, why should this clearly conservative approach be sufficient?

5.1 The moat is wide enough

The seller's consensus provided a clear answer. A Third Bridge expert condensed it into one sentence:

Stablecoins challenge the economics of the four-party model, not its infrastructure—they lack consumer protection, fraud management, and trust layers, and it is precisely this gap where Visa and Mastercard hold their strongest position.

Why is consumer protection a moat? Because it’s expensive. A single chargeback requires a full process of evidence submission, investigation, adjudication, and fund reversal, involving four parties: the cardholder, the merchant, the issuing bank, and the acquiring bank—each step demands manual intervention. Visa has refined this system over decades, accumulating precedents, deadlines, evidence standards, and appeal mechanisms. Others aren’t incapable of building it—they simply can’t justify the astronomical cost of rebuilding it from scratch.

The moat is wide enough, but the portion of goods with zero dispute has already moved off the track—such as buying tokens, calling APIs, or renting computing power in A2A machine transactions—where delivery equals consumption, leaving no room for reversal. On-chain transactions have no authorization, no recourse, and no dispute channel; for these goods, "naked" is precisely the advantage.

They left not because it was cheaper on-chain (though it was), but because they never used the feature provided by the card. Conversely, those still on the card today are precisely the ones for whom the cost of dispute resolution hasn't come down yet.

So the question isn’t whether stablecoins will become cheaper—it’s where the cost curve for dispute resolution will settle. If an agent can automatically retrieve order records, shipping proofs, product photos, and chat logs; cross-reference the merchant’s refund policy; auto-generate dispute filings; and follow them through to resolution, the marginal cost of handling a single dispute could drop from “tens of dollars in labor” to “a fraction of a cent in computation.” At that point, the system and its rules remain, but the cost barrier that supports them will thin significantly.

This is a bet that the cost of dispute resolution will decline, not that the system itself will fail. If you’re wrong, Third Bridge’s analysis is correct, and Visa’s position is even more secure than it appears.

5.2 Clearly state the weight assumptions

All the previous analysis—the two paths, the four tolls, the 29 basis points—rests on an unstated assumption.

A stablecoin payment might not require any card network at all?

If merchants collect digital dollars directly from customers' or businesses' wallets, Visa isn't paid less—it's simply not involved. There won't be a smaller number on the income statement; there will be no number at all.

This scenario is currently being piloted in Japan. Starting January 26 of this year, stores such as "Edo Shokuhin-kan" at Terminal 3 of Haneda Airport began accepting USDC payments in a real-world pilot, initiated by NETSTARS, the operator of the payment gateway StarPay, in collaboration with Japan Airport Terminal. The initiative targets inbound tourists, allowing users to scan store QR codes and pay directly using USDC from their personal wallets like MetaMask. Previously, SBI VC Trade and Aplus announced plans to launch a similar merchant payment pilot in spring 2026, while Lawson is testing a solution that uses existing POS systems without dedicated terminals for stablecoin settlements.

So why didn’t it happen on a large scale today? Four things are holding it back: habits, rewards funded by transaction fees, dispute rights, and the voucher already in your pocket. Break them down—one by one, none are impossible to replicate: rewards can be funded by wallets or merchants themselves, dispute mechanisms can be rebuilt on new platforms, and vouchers can be integrated into apps. The only thing holding it all together is habit.

In regions where such practices are not customary, leaks have already begun—cross-border B2B, drop shipping—these traffic streams involve no cardholders, no rewards, and no chargeback expectations; the advantage of stablecoins lies purely in cost efficiency.

Visa's stablecoin strategy appears relaxed because the segment it focuses on happens to be the most entrenched. But habits are not eternal.

In the end, there's still one question left unanswered: If neither stablecoins nor agentic systems are at the toll booth, what exactly are investors paying the premium multiple to Visa for?

Reviewing the sell-side commentary from this year, the answers appear with such frequency that they’ve become somewhat tedious.

For the January earnings report, McInerney described the quarter as “very strong,” with revenue and EPS both up 15%, attributed to resilient consumer spending, a robust holiday season, and growth in value-added services and business products. The April earnings report was headline-driven by “Visa’s strong performance fueled by resilient consumer spending.” For the July quarter, the opening narrative remained focused on continued strength in consumer and commercial spending, despite macroeconomic uncertainty.

The issue was stated most plainly during the March stock price correction. The debate at the time was summarized as: payment stocks are being reevaluated based on whether stable consumer spending can continue to support premium valuations. To address the concerns, Visa sent its Chief Product and Strategy Officer, Forestell, to Wolfe’s FinTech Forum, where he presented data showing an 8% increase in U.S. payment volume in January—9% for credit and 6% for debit.

A company called "Future Payment Infrastructure" offered as its answer to declining stock prices the number of cards Americans swiped in January. This is the anchor for Visa’s valuation.

Now consider the second anchor: portfolio migration. Last quarter, value-added services grew at a fixed exchange rate by 27%, accounting for 30% of net revenue; this quarter, they reached 34%, nearing one-third. The sell-side growth model is essentially “consumer payments provide the base, while value-added services and commercial payments provide acceleration.” The third anchor is buybacks—$20 billion in new authorizations in April, with an additional $4.9 billion repurchased this quarter.

Three anchors, none of which is called a stablecoin.

What are stablecoins and agentic in the eyes of sell-side analysts? The most typical phrasing is: "Visa's bullish case includes healthy consumer spending, stable cross-border activity, expanding value-added services, and growing blockchain initiatives."

The first three items have numbers; the last one is an adjective hanging at the end of the sentence. It’s not absent—it’s there, but it doesn’t carry weight.

So the full picture is this: the market is buying current global consumption, plus a future that doesn't need to be priced.

The former includes numbers, quarterly verification, and a model; the latter only requires existence—only proof that the company did not miss the next technological cycle.

Visa provides exactly this—it has built a ramp at every possible junction where one might bypass it, but each ramp is narrow. On the issuance side, it secures a seat at the alliance table; on the settlement side, it runs pilots across nine blockchains; on the operations side, it implements VSP; on the card processing side, it integrates tokenized deposits via Pismo; and on the front end, it extends its token infrastructure to agents.

The density of the ramp and the priority of investment are clearly mismatched, and this mismatch itself reveals its nature: it is insurance and options, not a bet.

Buy insurance and options early and broadly, but not at a high cost.

Go back to the beginning. That headline that flooded our feeds was Visa’s fourth-tier investment. The disconnect was never between “what Visa said” and “Visa’s finances”—every word McInerney uttered was measured; we were the ones who overinterpreted it.

Inflation is happening on the narrative side, not the company side.