Visa

IThe business

Visa runs the network that connects a card, a store and two banks. It authorises, clears and settles payments between the bank that issued the card and the bank that serves the seller. It does not issue cards, lend money or set the fees cardholders pay.² About 14,500 financial institutions use it. Visa counted 5.2 billion cards in the June 2026 quarter, and they can be used at more than 175 million merchant locations in over 200 countries and territories.²⁵⁰ In FY2026 it processed 281.3 billion transactions, up 9%.⁴⁹ Total payments and cash volume for FY2025 was $17 trillion.²

Net revenue grew from $23.0B in FY2019 to $40.0B in FY2025, 9.7% a year, and was $45.9B in FY2026, up 14.8% C (calculated).⁷²⁴⁹ The cash costs are small: capital spending was 3.6% of net revenue and stock pay 2.0% in FY2026 C (calculated).⁴⁹

How Visa is paid

Visa has four revenue lines and one deduction. Service revenue is a fee on payments volume, assessed on the prior quarter’s volume. Data processing revenue is a fee on transactions. International transaction revenue is earned on cross-border payments and currency conversion. Other revenue holds licences and services.² Client incentives are payments to banks, sellers and partners to win and keep volume, and Visa deducts them from revenue.² Value-added services, such as fraud tools, issuer processing and consulting, are not a separate line: they sit inside service, data processing and other revenue, before incentives.²

What Visa earns on each $100

Net revenue per $100 of payments volume rose from 24.2 cents in FY2021 to 29.8 cents in the TTM C (calculated).¹³²³ FY2021 is a poor start line: cross-border travel had collapsed, and international transaction revenue was $6.5B, against $7.8B in FY2019.¹³⁷ From FY2023 the figure can be split, because Visa discloses the revenue of value-added services: $7.2B in FY2023, $8.8B in FY2024, $10.9B in FY2025 and $13.4B in the TTM C (calculated).²³

The per-$100 figures in this part stay at the twelve months to June 2026. They need Visa’s payments volume in dollars and its value-added services for the full year, and both arrive with the FY2026 10-K, which is not filed yet. The fourth quarter did not change their direction: client incentives grew 21% in the quarter against about 15% for revenue before incentives C (calculated).⁴⁹

Visa does not say which revenue its incentives are paid against, or how much of them buys services volume. The split below therefore has two readings. The first charges every incentive dollar to the core network, which is the harshest reading and the one my cases use for their driver structure. The second shares incentives in proportion to revenue before incentives, 27.4% in FY2023 and 28.2% in the TTM to June 2026.²³

On the first reading, take value-added services out and the picture changes. Net revenue per $100 of volume rose from 27.0 to 29.8 cents between FY2023 and the TTM. Value-added services added 3.0 cents before incentives and the core after incentives gave back 0.3. The core network, meaning everything else, charged 31.2 cents before incentives in FY2023 and 32.5 in the TTM, paid out 10.2 cents in incentives and then 11.7, and kept 21.1 and then 20.8 C (calculated).²³¹⁴¹⁵

On the second reading the core after incentives rises from 22.7 to 23.3 cents, and services keep 6.4 of their 9.0, up from 4.3 of 6.0 C (calculated).

On either reading value-added services supplied most of the rise in net revenue per $100 of volume: 3.0 cents before incentives, and 2.1 after their share of them, against a total rise of 2.8. What changes is the core. It gave back 0.3 cents on the first reading and gained 0.6 on the second. Two facts do not depend on the split. Net revenue per $100 of volume rose 2.8 cents. Incentives rose 42% and revenue before incentives 38%, so incentives went from 27.4% to 28.2% of that revenue C (calculated).²³

Over those 2.75 years payments volume grew 24%, net revenue 36% and net revenue without value-added services 22% on the first reading C (calculated). Value-added services were 52% of the rise in net revenue if none of the incentives belong to them, and 37% if they bear their share.²³ The core yield before incentives rose 1.4% a year and incentives per dollar of volume 5.1% a year, so on the first reading the core yield after incentives fell 0.4% a year C (calculated).

Services supplied most of the rise in net revenue per $100 of volume; whether clients took the rise in the core yield depends on how Visa splits its incentives, which it does not say, and on the harshest split they took more than all of it.

Two cautions apply to the split. First, part of value-added services are Network Products, such as account verification and Account Updater, which Visa sells to sellers and acquirers to enhance transactions on its network.² Those are fees on the core network that Visa books as services, so the split overstates how separate the services are and understates the core yield. That makes it an upper bound on services and, on the first reading, a lower bound on the core. Second, core revenue before incentives per dollar of volume is a yield and not a price. It includes cross-border revenue, which grew 6% in the June 2026 quarter against cross-border volume up 12% excluding intra-Europe, because of lower currency volatility and mix, and processing revenue that follows transactions, not dollars.³⁴¹⁶ Over FY2023 to FY2025, before that effect, the core after incentives went from 21.1 to 20.9 basis points on the first reading and from 22.7 to 23.2 on the second C (calculated).²¹⁴¹⁵ A squeeze on the core would reach the services sold on it, so in my bear case the two fall together.

Three drivers set my valuation: volume, incentives against core yield, and value-added services. The incentive assumption does not depend on the split, because it is set on total incentives. The base case takes incentives from 28.5% of revenue before incentives in FY2026 to 34.1% in year 15, the bear case to 40.1% and the bull case to 28.4% E (estimate).²⁴⁹

Who pays, and where

Visa earns net revenue primarily from issuers and acquirers. One client was 11% of net revenue in each of FY2023, FY2024 and FY2025, about $4.4B on FY2025 net revenue C (calculated).² The United States was 38% of net revenue in the twelve months to June 2026 and international markets 62%; the split for FY2026 comes with the 10-K, and no other country reached 10% in FY2025.²³ Visa reports one segment and discloses no profit by region. International net revenue grew 16.5% in those twelve months and U.S. net revenue 11.1% C (calculated).³¹⁷

Cycle or utility

Card volume tracks consumer spending, so Visa has some cycle in it. In FY2020, with cross-border travel shut, net revenue fell 4.9% and operating income 6.1% C (calculated). International transaction revenue fell 19.3% and payments volume still grew 1.1% C (calculated).¹⁸¹³⁷ Net revenue then grew 10.3% in FY2021.¹³

The runway is the shift from cash and cheques. In the twelve months to June 2025 cash volume was $2.5 trillion of $16.4 trillion of total volume, 15% C (calculated). Consumer credit was $5.6 trillion, consumer debit $6.6 trillion and commercial $1.7 trillion; debit grew 9% in nominal terms and credit and commercial 5%.² Visa Direct transactions rose 21% to 4 billion in the June 2026 quarter, and commercial payments volume rose 13% in constant dollars.¹⁶

IIThe moat

I think Visa’s moat is a network effect, reinforced by scale in processing and by the switching costs of banks that have built on it. It is wide. On the evidence below it is stable in the core and not widening against Mastercard, and the rise in the core yield may have gone to clients.

A card is worth carrying if sellers take it, and a seller takes it if cardholders carry it. Visa counted 5.2 billion cards in the June 2026 quarter and has more than 175 million merchant locations.⁵⁰² On the Nilson count for calendar 2024 it had 4,805 million cards and 311 billion transactions against Mastercard’s 3,146 million and 204 billion.² A rival has to bring issuers, acquirers, sellers and cardholders at once. Capital One’s Discover network is the one live test: Capital One owns it and is moving its cards onto it in phases, debit first. Neither company gives the volume at stake, so it cannot be sized.²¹⁹

Does the moat hold share?

Each 10-K prints the Nilson Report chart for an earlier calendar year, so share can be followed over time. Visa carried 58.6% of the five networks’ payments volume in 2018, 58.9% in 2020, 57.4% in 2022 and 56.5% in 2024. Mastercard’s share rose from 30.1% to 33.7% C (calculated).⁷¹³¹⁵² Visa’s lead over Mastercard narrowed from 1.95 times to 1.68 times, because Visa’s payments volume grew 8.0% a year over those six years and Mastercard’s 10.8%. UnionPay and local networks are outside the chart.

The moat keeps volume growing. It has not kept share against the one rival that sells to the same banks and sellers.

What the record shows

The evidence for the moat is volume that keeps growing, a net revenue line that has risen 36% in 2.75 years, and an operating margin on Visa’s own measure that has stayed between 64.7% and 68.0% since FY2019 C (calculated).³⁸¹⁴¹³ The evidence against it is in the contracts. Visa’s 10-K warns that, in certain circumstances, clients may leave:

on relatively short notice without paying significant early termination fees

Visa pays to keep volume. Client incentives were 18.4% of revenue before incentives in FY2016 and 28.5% in FY2026. In FY2021 Visa described incentive contracts of less than one to fifteen years, and it expected by the end of FY2026 to have renewed incentive contracts covering about 20% of payments volume.^20491316 Each renewal is a negotiation in which the bank can say what it would do otherwise. If every incentive dollar belongs to the core, clients have taken more than all of the rise in the core yield over the last 2.75 years; if incentives are shared in proportion to revenue, they have not.

Where reach is still growing

Europe. Credentials in Europe grew by more than 40 million in twelve months, over 70% faster than the FY2019 to FY2024 pace, with 30 million more expected from wins already signed.¹⁶

Tokens and security. Card-not-present token penetration was 54.1% in FY2025 against a goal of 52.6%, and nearing 60% of e-commerce transactions by mid-2026.¹⁰¹⁶ A token ties a card to a device or a seller through Visa’s systems. Visa gives no client-retention or switching figure for tokens, so their effect on the moat is untested.

Services, growth outside the moat. Value-added services grew from $7.2B in FY2023 to $13.4B in the twelve months to June 2026, and were 21.6% of revenue before incentives against 16.0% C (calculated).²³ Visa sizes the annual revenue opportunity in value-added services at about $520 billion. The figure is Visa’s own.² The services are offered with the network and also as standalone products, and Visa lists technology companies, consultancies and merchant services firms as rivals in them, so they face competing offers.²

Evidence it is narrowing

Share. Visa’s share of the five networks slipped from 58.6% to 56.5% in six years, as above.⁷²

Incentives. They rose 42% in 2.75 years while revenue before incentives rose 38% C (calculated).²³ Clients capture part of the growth in what Visa earns.

The edges. Real-time payment networks have launched in at least 80 countries, including FedNow in the United States, Pix in Brazil and UPI in India.² UnionPay is the predominant domestic processor in China, and Visa’s licence application there has been pending since May 2020.²

The law. Interchange is paid by acquirers to issuers, and Visa generally receives none of it. It sets default rates, and its 10-K says that when it cannot set them at the optimal level, issuers and acquirers may find its network less attractive and issuers may cut card benefits.² The settlement cut of 10 basis points in average U.S. credit interchange, the Illinois statute and the Credit Card Competition Act therefore act on issuers and sellers.²¹²²²³ They reach Visa through three routes: the incentives issuers demand, the volume that moves when routing rules change, and the position of debit. The Justice Department’s debit case goes to that last route. The AP reported from the Justice Department’s announcement that Visa handles about 60% of U.S. debit transactions and collects over $7B a year in fees on them.²⁴

The next platform: agents and stablecoins

On the July 2026 call the CEO named stablecoins and agentic commerce as the two areas where Visa’s new way of building products is showing results.¹⁶ The CEO’s view of the second:

agentic commerce will expand our addressable market

Visa announced a partnership with OpenAI to enable Visa payments inside agentic commerce, and with Meta to enable payments across Facebook and Instagram on Visa Intelligent Commerce. It launched a Visa Stablecoin Platform and joined Open Standard, which plans to issue a stablecoin called Open USD.¹⁶ The 10-K describes Visa Intelligent Commerce as tokenised credentials with authentication for agent-driven transactions, a Trusted Agent Protocol that lets sellers verify agents, and a 2025 pilot that processed live agentic token transactions.²

The asset an agent needs is the token and the trust layer around it. When software enters the credential, someone has to say who issued it, whether the agent is authorised and whether the purchase reflects what the buyer wanted. The CEO named trust as the thing that will speed adoption.¹⁶ Visa holds the credentials and had tokens on nearly 60% of its e-commerce transactions by mid-2026, which is the starting position an agent platform needs.¹⁶

The bear route is that agents do not need cards. An agent platform can route payments over account-to-account rails or settle in stablecoins, or can set its own credential rules and negotiate with every network on price. Visa’s own 10-K lists stablecoin platforms and account-to-account rails among its competitors.² Stablecoin settlement through Visa was over $2.5 billion annualised at 30 September 2025, small against $17 trillion of volume, and Visa gave no updated figure in the June 2026 filings or on the call, where the CEO said stablecoins had yet to scale beyond a few use cases.²¹⁶ No revenue from agent payments or stablecoins is in any of my cases.

The moat is wide, it is not widening against Mastercard, and the next platform will test whether the token layer stays with Visa.

IIIThe competitive advantage

Visa’s advantage is scale. In calendar 2024 it carried 56.5% of the payments volume of the five networks in the Nilson chart in its 10-K, which leaves out UnionPay and local networks. Mastercard carried 33.7%, American Express 7.4%, JCB 1.3% and Diners Club and Discover 1.1% C (calculated).² Its payments volume was 1.7 times Mastercard’s, its cards 1.5 times and its transactions 1.5 times C (calculated).²

Scale has not bought faster growth. Mastercard’s net revenue grew 16% in calendar 2025 to $32.8B; Visa’s grew 11% in fiscal 2025 to $40.0B, and the two years end one quarter apart.⁹² In the June 2026 quarter, on the same dates, both grew 14%.²⁵⁴ Visa’s September quarter grew 13% and its full FY2026 15%, to $45.9B.⁴⁹

Scale has bought a higher margin. Visa’s GAAP operating margin was 60.0% in FY2025 against 57.6% for Mastercard in 2025. Before litigation, severance and acquisition items it was 67.7% against 59.2%. In the June 2026 quarter Visa’s GAAP margin was 59.1%, after $563M of severance, against Mastercard’s 60.2%; before those items it was 66.7% against 61.1%.²⁹⁴²⁵ Mastercard’s net revenue was 30.8 basis points of its gross dollar volume in 2025 and Visa’s 28.8 of its payments volume, on different definitions; Mastercard’s volume excludes Maestro and Cirrus, which raises its yield.⁹²

The market prices the two alike. The Mastercard figures here are its close of $623.60 before Visa reported and its earnings for the twelve months to June 2026: 34.3 times GAAP earnings and 33.2 times adjusted earnings, against 33.3 and 30.9 times FY2026 earnings for Visa C (calculated).^269255149 I have not rolled the Mastercard side forward for its latest quarter. The two networks earn high margins on a similar structure, which is evidence of how wide a moat they share.

Where Visa wins

Acceptance. Visa is accepted at more than 175 million merchant locations, and its lead over Mastercard in cards and transactions is 1.5 times.²

Processing. Visa processes transactions for other networks too, through gateway routing, and sells issuer processing through Visa DPS and Pismo.²

Growth markets. In Colombia, Visa’s processing penetration is above 90%, from single digits five years ago.¹⁶

Where it is exposed

U.S. debit. The Justice Department’s case is aimed at Visa’s strongest position and seeks injunctive relief.² A remedy would change how debit transactions are routed, and Visa would then compete for volume it holds today.

Credit cards in the United States. The 10 November 2025 settlement, preliminarily approved on 9 June 2026, lets sellers decline premium and commercial cards, widens surcharging, and cuts average credit interchange by 10 basis points for five years.²¹²⁷ Issuers lose interchange and are likely to ask Visa for more incentives on the volume that remains.

Alternatives. The 10-K lists American Express, Discover, JCB, Mastercard and UnionPay as global networks, local networks in many countries, closed ecosystems, instalment lenders, crypto and stablecoin platforms, and account-to-account rails.² Visa gives no volume figures for them, and the Nilson chart in its filing leaves UnionPay and local networks out.²

IVManagement

Ryan McInerney has been Visa’s chief executive since 1 February 2023. He joined in 2013 as president, after leading consumer banking at JPMorgan Chase from 2010 and consulting at McKinsey from 1997 to 2005. He was 50 at the December 2025 proxy.¹⁰ Al Kelly was chief executive from December 2016 and executive chairman until January 2024. Charles Scharf, his predecessor, ran Visa when it bought Visa Europe in 2016.²⁸¹⁰²⁹

Record

Net revenue grew 11.4% in FY2023, 10.0% in FY2024, 11.3% in FY2025 and 14.8% in FY2026, 11.9% a year from FY2022, against 10.4% a year from FY2019 to FY2026 C (calculated).^1411749 From FY2022, the last full year before McInerney became chief executive, to FY2026, net revenue rose from $29.3B to $45.9B, up 57%, and non-GAAP earnings per share from $7.50 to $13.27, up 77% C (calculated).¹¹⁴⁹ FY2022 growth of 22% was a rebound from the pandemic, growth held near 11% for the three years after it, and FY2026 stepped up to 15%. The 77% splits into revenue, margin, non-operating items, tax and share count:

Revenue growth did most of the work, and the lower share count added 12%. The tax rate and the margin did not help on Visa’s non-GAAP basis. Net revenue growth was 12% in the fourth quarter of FY2025, then 15%, 17%, 14% and 13%.^83031449

Decisions made on his watch include the renewal of incentive contracts covering about 20% of payments volume by the end of FY2026, the Europe wins that added more than 40 million credentials in a year, the $1.5B purchase of Prisma and Newpay in Argentina, the pending $2.4B purchase of BioCatch announced in August 2026, and a severance programme.¹⁶³⁴⁹ In July 2026 Visa said it is eliminating roles, mostly in technology and product, and booked $563M of severance in the June quarter. It said it would reinvest the savings and expects strong margins.¹⁶⁴ Management is cutting cost while growth is strong.

Pay

McInerney’s pay was $31.6M in FY2025, up from $26.0M and $22.6M, and 204 times the median employee’s $154,909.¹⁰ Of it, $18.7M was stock, $5.7M options and $5.3M an annual cash incentive paid at 140% of target.¹⁰ The incentive pays on net revenue growth, net income growth and earnings per share growth, with business drivers beside them. On the plan’s adjusted basis, net revenue grew 11.2% in FY2025 against a goal of 9.8% and earnings per share 15.1% against 12.2%, while payments transactions grew 9.3%, under a goal of 10.4%. Reported non-GAAP earnings per share grew 14.1% and GAAP earnings per share 4.8%.¹⁰⁸² His equity is 50% performance shares, earned on earnings per share goals with a three-year relative return modifier against the S&P 500, 25% options and 25% restricted stock units.¹⁰

The plan rewards growth in net revenue and earnings, which counts the services mix as much as the core fee, and its earnings measure leaves out items Visa treats as non-recurring, litigation provisions among them.¹⁰ Five-year total return to 30 September 2025 turned $100 into $176.94 for Visa and $245.77 for the S&P 500 Financials index, the peer group in the proxy, 68.83 points behind C (calculated).¹⁰ Say-on-pay support was 93.28% in January 2026.³²

Ownership and sales

At 1 December 2025 McInerney held 274,569 shares, about $113M at the Q4 2026 price, and 823,086 including options he can exercise within 60 days. His guideline is six times salary, about $9M C (calculated).¹⁰⁵¹ A later total cannot be computed from the Form 4 filings. Between October 2025 and October 2026 he sold 122,480 shares for about $42.3M, each time alongside option exercises and under 10b5-1 plans. Six officers sold about $89M together, and none bought in the open market C (calculated).¹² Some sales were outside 10b5-1 plans, for example Suh’s on 12 May 2026 and Mahon Tullier’s on 30 July, each soon after earnings.¹² The pattern is regular selling by officers whose pay is mostly in stock.

The bench and the board

Chris Suh has been chief financial officer since July 2023, after Electronic Arts and more than 25 years at Microsoft. Rajat Taneja has run technology since 2013, Kelly Mahon Tullier joined in 2014 and is vice chair, and Paul Fabara has led risk and client services since 2019.¹⁰ John Lundgren has been independent chair since January 2024. The board has eleven nominees, ten independent plus the CEO, with an average tenure of 6.3 years.¹⁰²⁹ An independent-chair proposal got 16.08% support in January 2026, and only class A holders voted.³² Visa set Delaware courts as the exclusive forum for certain claims against it in July 2026.³³

Capital allocation

Visa’s stated order is to invest in the business first, pay 20% to 25% of earnings per share as dividends and return the rest through buybacks.³⁴ The record has one large deal, the 2016 purchase of Visa Europe, one blocked deal, Plaid, and a run of small ones. The buybacks have been the large decision.

Since FY2016 Visa has spent $105.4B on 595 million shares, an average of $177 a share, and diluted class A shares fell 18.6% over those nine years, from 2,414 million to 1,966 million C (calculated).²⁰² At $410.54 those shares are worth $244B, $139B more than they cost C (calculated).⁵¹ The shares bought in FY2025 cost an average of $335.44 and those bought in FY2026 $335.49, 4% under my base value of $349; the 11 million bought in the fourth quarter cost $369.12 C (calculated).²⁴⁹ Visa paid $5.1B in dividends in FY2026 and $20.5B for stock, 103% of owner free cash flow C (calculated).⁴⁹ The dividend has grown 15.3% a year since FY2019 and is set at 20% to 25% of earnings per share.³⁴⁴⁹

The test for a buyback is price paid against value at the time, which cannot be rebuilt year by year here. On outcome the FY2016 to FY2025 record is good, and the FY2026 purchases were a little under my base value. The next ones are not: at $410.54 each $1 of buyback buys about 85 cents of my base value, 89 cents of the like-for-like base, and the $24.4B of authorisation left at 30 September 2026 would transfer about $4B to sellers E (estimate).⁴⁹ The performance shares are earned on earnings per share, which a buyback lifts at any price; the relative return modifier is the only price test in the plan.¹⁰

The deals since 2022 are small: Currencycloud $893M, Tink $1.9B, Pismo $929M, Featurespace $946M, and Prisma and Newpay in Argentina for $1.5B.¹¹¹⁴²³ Together they cost $6.2B, against the $20.5B Visa spent on its own stock in FY2026, and the pending BioCatch deal for $2.4B would be the largest of them C (calculated).⁴⁹ Goodwill and acquired intangibles are $48.2B against equity of $36.1B, and they stood at $42.3B at 30 September 2016, after the Visa Europe purchase C (calculated).⁴⁹²⁰ The Plaid deal, agreed for $5.3B in January 2020, ended after the Justice Department sued to block it.¹⁸¹³

Follow the pay. Half of McInerney’s equity is earned on earnings per share, and a buyback lifts that number at any price, so the plan pays him to shrink the share count even when the stock is dear.¹⁰ His stake is about $113M, but he sold about $42.3M in a year and no officer bought.¹⁰¹² The nine-year buyback record is good. The next ones are the test.

The verdict

Execution is strong, pay is mostly stock and is paid on growth rather than on return, and the buyback record turns on price.

VFinancials

Visa is a high-margin business whose profit is partly hidden by litigation provisions and whose cash flow is partly hidden by legal payments.

GAAP operating margin held between 64% and 66% from FY2019 to FY2024, then was 60.0% in FY2025 and 61.2% in FY2026, after litigation provisions of $2.6B and $1.8B C (calculated).⁷²⁴⁹ On Visa’s own measure, before litigation, severance and acquisition items, margin was 67.7% in FY2025 and 67.5% in FY2026, when it also left out the $282M Visa Foundation donation.²⁴⁹

Earnings quality

Visa’s non-GAAP measure leaves out nearly all litigation provisions, $1.72B of the $1.76B booked in FY2026, so the covered and the uncovered provisions are taken separately. In FY2026, provisions of $1.8B came from the interchange class action and its opt-outs and from matters outside the plan.⁴⁹ The uncovered provision, which I carry at 0.9% of net revenue, $0.41B, as in the twelve months to June ($397M), is kept as a cost; the FY2026 split comes with the 10-K, and the covered provision is left out because the class B pool absorbs it, a mechanism explained under the valuation C (calculated).²³

FY2026 net income of $23.5B and GAAP earnings per share of $12.32 include a net deferred tax benefit of $314M from a change in the U.S. taxation of certain foreign earnings, which Visa excludes from its non-GAAP earnings per share of $13.27.⁴⁹ The FY2026 tax rate of 16.3% would be about 17.5% without it C (calculated).⁴⁹

Cash

Free cash flow, as Visa counts it, was $22.1B in FY2026, 48% of net revenue.⁴⁹ It converts at close to net income: 108% in FY2025 and 94% in FY2026 C (calculated).²⁴⁹ The FY2026 ratio is lower because Visa paid about $3.8B for legal matters, $3.0B of it in the first nine months for U.S. covered litigation and mostly out of the escrow, against $1.3B a year earlier C (calculated).⁴⁹³ Payments run through operating cash flow. The deposits that fund the escrow do not, because restricted cash sits inside the cash-flow total.⁷⁴⁹ Adding back covered payments, about $3.7B by my estimate, and deducting stock pay gives owner free cash flow of $24.8B, the starting point of my model. On the same definition it was $12.4B in FY2019, $14.0B in FY2021 and $21.8B in FY2025 C (calculated).⁷¹³²

The balance sheet

Debt is $22.1B at carrying value against $14.2B of cash and investment securities, a net debt of $7.9B, 1.0% of enterprise value C (calculated).⁴⁹ Goodwill and intangibles of $48.2B exceed equity of $36.1B, so tangible equity is negative.⁴⁹ Visa had $9.2B of liquidity for daily settlement at 30 September 2025. It indemnifies clients if settlement fails: the maximum daily exposure was $168.6B and the average in the nine months to June 2026 was $99.5B, against $9.5B of collateral.²³ That is a tail risk the cash flows do not show.

VIValuation

First, how this goes badly wrong

Before any numbers, I worked back from what would hurt an owner most. Three things can.

• The banks take more of each dollar. Client incentives went from 18.4% of revenue before incentives in FY2016 to 28.5% in FY2026, and more volume renews in FY2027.²⁰⁴⁹⁵⁵ At the record’s pace, 5.1% a year per dollar of volume, my base value is $299, not $349 E (estimate).
• The debit rules change. Visa handles about 60% of U.S. debit and collects over $7B a year on it, per the Justice Department.²⁴ A remedy that opens routing is on my list of what lowers my value.
• Payments find another route. Real-time networks run in at least 80 countries, and Visa’s 10-K names stablecoins and account-to-account rails as rivals.² None of that is in my cases.

What an owner can take out is close to what Visa reports, because the business needs little capital. Free cash flow was $22.1B after $1.7B of capital spending, and stock pay took $0.9B.⁴⁹ That is $11.30 a share. Adding back the $3.7B of covered lawsuit payments, which the class B pool pays, gives the $13.28 my model starts from C (calculated). Earnings were $12.32 GAAP and $13.27 non-GAAP. At the price, owner cash yields 2.8% to 3.2%, under the 5.00% Treasury.

I cannot know how fast incentives rise, whether services stay above 20% as the business doubles, or whether the class B pool lasts as long as the claims. So I give a band. Incentives alone run from $299 to $380, services alone from $298 to $409. The band I can defend is about $300 to $440, and $349 is one point in it, not a measurement E (estimate).

The price needs services on the bull path and incentives back at my 2.8%, when they run near 5.5%. I put each at about one in three, so both is about one in nine, and nearer one in six, since new deals bring services E (estimate).

What the price implies

At $410.54 Visa trades at 33.3 times its GAAP earnings and 30.9 times its non-GAAP earnings for FY2026, and at 27.3 times consensus non-GAAP earnings for FY2027, which is vendor data retrieved in Q4 2026 E (estimate).⁵¹⁴⁹⁵³ Owner free cash flow is 3.2% of enterprise value of $774.6B C (calculated).⁵¹⁴⁹ The close is 1.0% under the 52-week high of $414.60 set five sessions before the Q4 2026 close and 14.3% above the 30 September close of $359.33 C (calculated).⁵¹⁵²

From $410.54 my bear value is 51% lower, my weighted value of $353 is 14% lower, my base value is 15% lower and my bull value is 24% higher E (estimate).⁵¹

Then the rate. At $410.54, my cash flows are worth the price at a discount rate of 5.4% in the bear case, 8.2% in the base case and 10.1% in the bull case, against a 5.00% ten-year Treasury yield and the 9.0% I require E (estimate).⁵¹⁶ That implied rate values year 15 at the same rate. If year 15 is valued at 9.0% instead, my cash flows return 2.2%, 7.5% and 11.0% a year at the price, and the bear case returns less than the Treasury E (estimate).

Solved as one rate, the price needs owner free cash flow to grow 8.3% a year for 15 years, from $24.8B to about $82B in year 15, with growth of 3.5% after that C (calculated). With growth after year 15 of 3.0% the rate is 8.7%, and with 4.0% it is 7.8%. On the same definition owner free cash flow grew 10.4% a year from $12.4B in FY2019 to $24.8B in FY2026, and 11.7% from FY2021 to FY2025 C (calculated).⁷¹³²⁴⁹ The price asks for less growth than Visa delivered over the last 7 years, held for 15. In net revenue the price needs about 8.7% a year at my base case’s conversion into free cash flow, against 6.8% compounded in my base case, which is worth a steady 7.2%, 9.7% a year from FY2019 to FY2025 and 14.8% in FY2026 E (estimate).

At $410.54 the price needs a steady 8.3% growth in owner free cash flow for 15 years; my base path, 6.4% compounded, is worth a steady 6.8%, and my bull path, 9.3% compounded, a steady 10.2%.

The discount rate

I use 9.0%, the rate on my Microsoft page, as a hurdle for large, established companies. It is not a Visa-specific premium. The capital asset pricing model gives less for Visa. With a 5.00% ten-year Treasury yield and Damodaran’s 4.20% implied equity risk premium, Visa’s beta against the S&P 500 total return index gives 8.71% on ten years of monthly returns (beta 0.88), 8.24% on five years (0.77) and 7.30% on three years (0.55) C (calculated).^6363738 Litigation is not in the rate: the cash flows carry 0.9% of net revenue a year for matters outside the plan, and the bear case a $10B loss. The rate decides the verdict, so the table gives three: my base value is $409 at the five-year-beta rate, with the price level with it, $370 at the ten-year rate, with the price 11% above it, and $349 at 9.0%, with the price 18% above it E (estimate). The ten-year yield closed at 4.98% on 29 October, 2 basis points under the 5.00% used here, and I have not re-struck the three rates.⁵⁴

Dates and timing

My cases discount each year’s cash flow at year end and date the value to 30 September 2026, the last balance sheet. The price is the Q4 2026 close a month later. Both conventions lower the value. Mid-year timing adds $16 to the base value, dating it to the price date adds $2, and together they add $18, which takes the base to $367 and puts the price 12% above it. The weighted value goes from $353 to $371, the base-case implied rate from 8.2% to 8.4%, and the growth the price needs from 8.3% to 7.8% E (estimate).⁵¹ The headline keeps year-end timing and the September date because both are conservative, and the like-for-like figures sit beside it.

Owner free cash flow and the litigation

The model starts from owner free cash flow, which is free cash flow after stock pay and before payments for covered litigation. The reason is in Visa’s capital structure. U.S. covered litigation, meaning the interchange class action and the merchants who opted out, is paid under a retrospective responsibility plan. When Visa deposits cash in the escrow, the conversion rates of its class B stock fall. Each deposit retires class A equivalents at the market price, which Visa says has the same effect on earnings per share as repurchasing class A stock.³⁹³ The cash leaves, the share count falls by the same value, and class A holders are where they were, for as long as the claims stay inside the pool and the market price equals value. About 93.6 million as-converted class B shares remain, worth $38.4B at $410.54.³⁹ The model keeps the as-converted share count of 1,868 million and leaves covered payments out of the cash flows. If the pool is used in full at today’s price, my base value falls about $3, because the shares retire above my value E (estimate).

The shield has limits. It covers U.S. covered litigation and the Europe preferred stock, and nothing else. The Justice Department’s debit case, the U.S. debit class actions and the securities and derivative suits fall on Visa and on class A holders directly. An amended consolidated complaint in the debit class actions covers Visa fees for debit routing services from 1 January 2012.²³ The securities suit was dismissed without leave to amend on 29 June 2026.³ Visa gives no loss estimate for the matters outside the plan, and says it cannot estimate a loss at trial for the interchange class action if talks fail.²³ The $17.4B of interchange at issue is Visa’s test for the class B exchange offer, not a damages figure; it was $39.4B in October 2025.⁴⁰⁴¹ The pool’s value moves with the price, so it is smaller if the stock falls while claims rise. It is large against the record: Visa provisioned $5.9B for all litigation in the ten years to FY2025 and $1.8B in FY2026, and the pool is worth about five times that C (calculated).²⁴⁹

Each $10B of loss that falls on class A holders costs about $4 a share after an 18.5% tax shield if it is paid today, and $3 if it is paid in year 3 C (calculated). My model carries 0.9% of net revenue a year for matters outside the plan, about $0.4B now, and my bear case adds a $10B loss in year 3. That is about one and a half years of the entire $7B a year in U.S. debit fees the Justice Department alleges, a severe case, because an overcharge is a fraction of the fees. It costs the bear case about $3 a share.²⁴

Three cases

Each case is built from payments volume down to free cash flow, over 15 years, because the network’s runway is long and a ten-year model would put most of the value in the terminal year. After year 15 comes 49% of base-case value, with growth of 3.5%. Growth after year 15 does not exceed the case’s net revenue growth in year 15, to one decimal: the bear case uses 1.7% against 1.7%, the base case 3.5% against 3.6% and the bull case 4.0% against 5.2%. Payments volume grows at a set rate. Core revenue before incentives per dollar of volume and incentives per dollar of volume each move at a set rate. Value-added services grow on their own path. Operating margin is Visa’s non-GAAP margin less the provision for matters outside the plan, tax is 18.5%, and net reinvestment, capital spending less depreciation, is 1.2% of revenue, against 1.1% in the twelve months to June.³ Acquisitions cost 1.0% of revenue a year: Visa spent $3.75B on them in the four fiscal years FY2022 to FY2025, 2.7% of revenue, then $0.7B in FY2026, and it has agreed to pay $2.4B for BioCatch.⁴⁹ The model assumes less than the old average because much of the growth in value-added services is organic, and if BioCatch came on top of the 1.0% my base value would be about $1 lower C (calculated).^1115142 Unvested stock awards are not added to the share count, because stock pay is already a cost in the cash flows. The cases are in nominal dollars with exchange rates held flat; exchange rates added about one point to net revenue growth in the June 2026 quarter.³

The base case carries the last three years forward at a slower pace.

• Payments volume grows 8.0% in year 1, below the 10% nominal growth of FY2026, and fades to 4.0% by year 15.⁴⁹
• Core revenue before incentives per dollar of volume rises 1.0% a year, below the 1.4% of the last 2.75 years, and incentives per dollar of volume rise 2.8% a year, against 5.1%. The core net yield then falls 0.3% a year, against 0.4% in the record.
• Value-added services grow 20% in year 1, from about 31% in FY2026 (my estimate), and fall to 4% by year 15. They reach $54B in year 15, 44% of net revenue and about 10% of the $520 billion opportunity Visa cites.²³
• Net revenue grows 11.7% in year 1, to $73B in year 5 and $123B in year 15, 6.8% a year. Visa guided FY2027 net revenue growth to low double digits, on an adjusted basis: non-GAAP, in constant dollars and excluding acquisitions, after growth of 15% in FY2026.⁴⁹⁵⁰
• Margin holds at 67.5% for five years and drifts to 66.0%, because management said it will reinvest the restructuring savings.¹⁶

The bear case is a regulatory and competitive squeeze. The Justice Department wins on debit and a routing law follows.²⁴ Issuers use the settlement and the threat of other networks to push incentives up 3.0% a year against core yield gains of 0.5%, so the core net yield falls to 16.5 basis points. Volume grows 7.5% in year 1 and fades to 3.0%, value-added services fade from 17% to 3%, margin falls to 63% and a $10B loss lands in year 3. Net revenue grows 3.6% a year.

The bull case is a world in which core yield outruns incentives. Core revenue before incentives rises 1.5% a year against incentives at 2.5%, the core net yield reaches 23.5 basis points, volume grows 9% falling to 4.5%, and value-added services compound at 26% falling to 5%, reaching $81B in year 15. Margin reaches 69.5%, and net revenue grows 9.3% a year. Weighted 25%, 50% and 25%, the three cases give $353 a share E (estimate), against a price of $410.54.⁵¹

What closes the gap

Two changes take my base value to the price. A discount rate of 8.24% gives $409, and value-added services on my bull path give $409 E (estimate). Incentives growing only as fast as the core yield, 1.0% a year, give $380, and together with the bull path for value-added services they give $440, above the price. The timing conventions close more of the gap than any one operating change: $16 for mid-year cash flows and $2 for the price date. The rate and the two Visa-specific drivers matter more than margin, tax or litigation: holding margin at 67.5% for 15 years adds $6 a share and a $10B loss in year 3 costs $3 E (estimate).⁵¹

The model is generous in one place and cautious in another. It assumes the class B pool outlasts the claims, and if it did not, my base value would fall by about $4 a share for each $10B of the shortfall E (estimate). It is cautious about working capital: Visa holds $12.2B of client incentive liabilities and settlement balances that grow with the business, and the model assumes no inflow from them.⁴⁹

The margin of safety

• Price$410.54Close, Q4 2026 [51]
• Base value$349My estimate E (estimate)
• Margin of safety below$26225% under my base value C (calculated)

I set the margin of safety 25% under my base value. That is a judgement and not a derived figure, so two case-based checks sit beside it. The price at which my bear value is 25% lower is $270. The price at which the bear case clears the 9.0% hurdle is the bear value itself, $203. The lower of the first and the 25% line, $262, is the margin-of-safety price, and the bear value is 23% lower there. At $262 my base cash flows are worth the price at a 10.7% discount rate and my bull case’s at 13.0% E (estimate). That is 19.7 times FY2026 non-GAAP earnings and 17.4 times FY2027 consensus.⁴⁹⁵³ The margin protects against the pace of value-added services and the price of incentives, where the cases differ most, more than against litigation. The price sits 57% above the margin-of-safety price.

Where this could be in 5 and 10 years

The value at each date is my model’s present value, at 9.0%, of the owner free cash flow after it, less $8.3B of net debt and accruals, per today’s 1,868M shares and in that year’s dollars. Owner cash is that flow from October 2026. The return counts price growth and that cash, exiting at my 2036 value E (estimate).

The base stands at five years if net revenue reaches $73.5B from $45.9B, the margin stays near 67.5%, and incentives grow 2.8% a year against 1.0% for the fee before them. It stands at ten if revenue reaches $100B, with services at $42.5B of it. Reinvestment and acquisitions stay at 2.2% of revenue. At 9.0% the value rises about 4% a year, and the rest of the return is the $212 a share paid out.

The bear is a squeeze. Incentives grow 3.0% a year against 0.5% for the fee, volume growth fades to 3%, the margin falls to 63% and a $10B loss lands in year 3. The bull is the reverse: the fee outruns incentives, services compound at 26% falling to 5% and the margin reaches 69.5%.

The most likely way my base is wrong is incentives. If they grow 4.3% a year instead of 2.8%, base value in 2036 is $464, not $527, and the return falls from 7.1% to 5.9%. Services come next: growth 2 points a year slower gives $473. At a 10% discount rate it is $446.

The next best use of the same dollar is Mastercard, the closest comparable on this site. Its base return is 7.6% a year against 7.1% here.

How good the business is and how good the price is are two questions. The business keeps a 67.5% non-GAAP margin on 3.6% of revenue in capital spending.⁴⁹ The price needs 8.3% growth in owner cash for 15 years. This looks like a strong business at a price that already counts on much of what services could deliver.

These are estimates of value, not forecasts of the share price. The market price can sit above or below value for years.

VIIThe case for the price

What most people miss

Most of the services line is tied to the network. On the October call (unofficial transcript), Ryan McInerney said transactions, cards and accounts drove more than 80% of value-added services revenue over the last five years.⁵⁵ So these sales grow when the network grows.

Tokens have barely been charged for. A token is a stand-in for your card number. Akre Capital’s Q3 letter, using Visa’s data, says more than half of global e-commerce transactions are tokenized, with 39% less fraud and a 5% lift in approvals. “Visa and Mastercard have only started monetizing this technology a decade after introduction.”⁵⁶

The bull case

Bill Ackman’s Pershing Square makes the strongest version. It started buying this year and reported 3.27M shares, worth $1.1B, at 30 June.⁶⁰ Its August report, on Visa and Mastercard together, calls both capital-light toll-takers, says cards are about half of the consumer spending they could win, and calls the fears over stablecoins, AI agents and U.S. regulation misplaced. Visa’s services are about 30% of revenue and grow two to three times as fast as payments. For Visa it estimated 23 times forward earnings at 30 June, and 16% yearly growth in earnings per share over three to five years.⁵⁷

McInerney put value-added services at about $14B in FY2026, up 30% in constant dollars, and expects the low to mid-20s in FY2027.⁵⁵ TD Cowen raised its target to $450 from $420 and said services can stay above 20%, as reported by Investing.com. Morgan Stanley went to $471 from $416, as reported by MarketBeat.⁵⁸⁵⁹

The filings are mixed. In the June quarter Akre cut 32%, to 1.11M shares, and Fundsmith cut 35%, to 2.02M.^61626364 TCI reported 30.5M, its second-largest position.⁶⁵ Berkshire Hathaway went from 8.30M shares at the end of 2025 to none at 31 March 2026, and gave no reason in the filing.⁶⁶⁶⁷

Against reality

Two things close the gap. Put value-added services on my bull path, 26% in FY2027 falling to 5%, and my base goes from $349 to $409. Slow incentives per dollar of volume to 1.0% a year and it goes to $380. Do both and it is $440 E (estimate). So if services follow that path and incentives grow no faster than the 2.8% in my base, the price is possible.

Services are close. Guidance, stated before acquisitions and currency, sits above my 20% and under the 26%. Fourth-quarter growth had help from FIFA work and the Prisma purchase.⁵⁵ The path still needs 15% growth in year 5, on a business more than twice today’s size E (estimate).

Incentives are far off. They rose 16% in FY2026, about 5.5% per dollar of volume on my estimate C (calculated), twice the 2.8% in my base, and 21% in the fourth quarter.⁴⁹ CFO Chris Suh said about 25% of volume renews in FY2027, against 20% in FY2026, and that incentive growth steps up.⁵⁵ My base has about 11%. If incentives rise 16% and net revenue 12%, they reach about 29% of revenue before incentives, not my 28.4%. At 5.1% per dollar of volume, the record’s pace, my base is $299 E (estimate). McInerney says the new deals also bring in services.⁵⁵

Pershing’s 23 times was at about $343 C (calculated).⁵⁷⁶⁰ At $410.54 Visa is 27.3 times FY2027 consensus, and FY2027 guidance for earnings per share is low double digits to low teens, under that 16%.⁵¹⁵³⁵⁰ McInerney called agent commerce very small for FY2027.⁵⁵ Cathie Wood, whose ARK bought Circle, posted on 23 August that analysts who built their records on Visa and Mastercard “cannot fathom Circle, the disrupter”, as reported by Mitrade.⁶⁸

Where I come down

The guidance moved me: services look stronger for FY2027 than my 20%, and that is the piece most likely to prove my base too low. Incentives cut the other way: Visa is paying more to renew, and each extra point a year takes about $20 off the base, so my $203, $349 and $509 stand. Services above 20% through FY2027, with incentives under 28.5% of revenue before incentives while volume grows 8%, would raise my values.

VIIIRisks, and what would change my view

1. Incentives. They were 28.5% of revenue before incentives in FY2026, and 29.7% in the fourth quarter alone, when they rose 21% against about 15% for revenue before incentives C (calculated).⁴⁹ They rose 42% in the 2.75 years to June 2026 against revenue before incentives up 38% C (calculated).²³ Each extra point of yearly incentive growth takes about $20 off my base value. A large client that moves volume, or a renewal that resets terms, would show up here first. On the October call (unofficial transcript) the CFO said about 25% of volume renews in FY2027, against 20% in FY2026, and that incentive growth steps up.⁵⁵
2. Client concentration. One client was 11% of net revenue in each of FY2023 to FY2025, about $4.4B on FY2025 net revenue, and the 10-K says clients may end contracts on short notice. If revenue followed volume, a move of 10% of that client’s volume would cost about 1% of net revenue C (calculated).² Any disclosed change in the 11% client is a signal.
3. The debit case and routing. The Justice Department sued on 24 September 2024 for monopolising debit. Its motion to dismiss was denied on 23 June 2025, and fact discovery is reported to run to 29 January 2027, with no trial date set.²⁴²⁴³ The relief sought is injunctive. The cost to Visa would fall on debit volume and pricing, and on damages in the follow-on class actions.
4. Value-added services. They supplied most of the growth in net yield. Visa lists technology companies, consultancies and merchant services firms as rivals, and says generative AI can create competing offers.² Growth was 24% in FY2025, about 31% in the twelve months to June 2026 and 33% in the June 2026 quarter. The release does not split value-added services out for the fourth quarter or the year. On the October call (unofficial transcript) Visa said 31% in constant dollars for the quarter, 30% for the year and the low to mid-20s for FY2027. The dollar figure comes with the 10-K; a drop below 15% is my signal.²³⁵⁵
5. Disintermediation by agent platforms and other rails. An agent platform could set its own credential rules, route payments over account-to-account rails or settle in stablecoins. Capital One is moving cards onto the Discover network with the volume at stake undisclosed. Real-time payment rails, stablecoins, wallets and national schemes are small against Visa now.²¹⁶¹⁹
6. Cross-border. International transaction revenue grew 8% in the September 2026 quarter against cross-border volume up 14%, a gap of 6 points, the same as in the June quarter (6% against 12%).⁴⁹⁴ In July the CFO put the June gap down to lapping last year’s peak in currency volatility and to mix, and assumed more of a drag from volatility in the fourth quarter.¹⁶
7. Payout above cash generation. Visa paid $25.6B in buybacks and dividends in FY2026, 103% of owner free cash flow, and carries $22.1B of debt against $14.2B of cash and investments C (calculated).⁴⁹ The pending $2.4B BioCatch purchase is cash and is expected to close by the end of fiscal Q2 2027.⁴⁹
8. Settlement. Visa indemnifies its clients if settlement fails, and its maximum daily exposure was $168.6B in the nine months to June 2026.³
9. Restructuring. Visa booked $563M of severance, cut product teams to squads of two to four, and has the Prisma and Newpay integration in Argentina under review by the competition authority.⁴¹⁶³
10. Cyber and technology. The 10-K names a failure or a breach of VisaNet as a risk.²

What would raise my base value: incentives growing no faster than price, value-added services above 20% through FY2027, and a debit case that ends without routing remedies. What would lower it: incentives above 30% of revenue before incentives, value-added services growth under 15%, or a remedy that opens debit routing.²³⁴²

The fourth quarter of FY2026 passed on payments volume, margin and the FY2027 outlook; incentives sat on the line, and cross-border revenue did not close the gap with volume. That confirms the thesis and I changed no case assumption.⁴⁹⁵⁰ Gaps in the record: how Visa splits its incentives between the core and services, the volume Capital One is moving, any stablecoin settlement figure after September 2025, FY2026 payments volume in dollars and value-added services, which I have estimated until the 10-K, the covered share of FY2026 legal payments, the share count at 30 September, which I rebuilt from the balance sheet, the like-for-like timing, which my headline values leave out, the Justice Department schedule, which rests on trade press and not on a court order, the signature on the Illinois law, the Regulation II appeal rulings, and Visa’s credit ratings, which are from February 2025.^424323451634 The next platform is not valued, and no growth rate here depends on it.