Visa (V) or American Express (AXP): two opposite models

2026-08-11 · By Lubin Danilo, founder of Lubin Investment

Visa never lends a single dollar to anyone: it moves information between banks and charges a tiny toll. American Express directly lends money to its own cardholders and carries the risk if one of them never pays it back. Same swipe at checkout, two almost opposite businesses, with very concrete consequences in the numbers.

Dinner with friends, the check arrives, two cards come out of two pockets. The first is an American Express card. The second is a Visa debit card issued by a small regional bank nobody has ever heard of. The terminal beeps twice, money leaves two accounts, everyone goes back to the conversation. Nothing tells the two payments apart: not the speed, not the screen, not the gesture.

And yet, in the seconds right after that beep, two almost opposite mechanics just kicked in. Visa never lent anything to anyone: it passes a message between the restaurant's bank and the customer's bank, collects a fraction of a cent along the way, and has never seen either person's face or credit file. American Express IS the cardholder's bank. It reviewed that person's file, set their credit limit, and if that person never paid their statement back, the hole would show up directly on American Express's own books.

Two identical payments for the customer, two publicly traded companies that get lumped into the same bucket ("card companies") even though they take almost none of the same risks. That is the mechanism I want to unpack here, invisible at the checkout counter but decisive on a balance sheet: who really stands behind a Visa transaction, why American Express works differently, and what that changes, very concretely, in the numbers I look at as an investor.

Four players at Visa, only three at American Express

A typical Visa or Mastercard transaction involves four distinct players. The cardholder, first. The issuing bank, next: it is the one that issued the card, set the credit limit, and actually lends the money if it is a credit card (Chase, Capital One, a regional bank, whichever: it is never Visa). The acquiring bank, on the merchant's side, which collects the payment on the merchant's behalf. And finally the network (Visa or Mastercard), which does exactly one thing: move the payment order between the two banks and guarantee the transaction is valid, in exchange for a tiny fee called an interchange fee, charged on every single swipe. Visa states this in black and white in its own annual report filed with the US regulator: the company does not issue cards and does not extend credit, as its fiscal year 2025 Form 10-K makes clear.

American Express was built the opposite way from the start: instead of four players, there are only three. American Express is SIMULTANEOUSLY the issuing bank, the network, and often the merchant's acquirer too. This architecture is called a closed-loop network, while Visa's is called open-loop, because it opens the game to thousands of third-party banks. No middleman to pay for American Express, but in exchange, it alone carries what Visa usually spreads across hundreds of different banks: the risk that a customer never pays back.

Why American Express lends money, and Visa never does

Every time an American Express cardholder pays and does not clear their statement that same month, the company just extended them a loan, funded by its own cash or by debt it borrowed on the markets itself. To cover the risk that some of those loans never get repaid, American Express sets aside an accounting reserve every quarter called a provision for credit losses: money frozen ahead of time, just in case. In the second quarter of 2026, that provision jumped from $1.02 billion a year earlier to $1.48 billion, according to the company's Form 10-Q filed with the US regulator. This is not an isolated blip: in the first quarter of 2026 already, the provision was up 9% year over year, while revenue was up only 11%, an almost matching pace. Credit is getting more expensive to set aside for, year after year.

This is not a sign of poor management either: American Express's 30-day delinquency rate stood at 1.4% in March 2026, a figure still below its own 20-year average of 1.5%, and among the best in its industry. But the trend speaks for itself: after several years of unusually low defaults, the trajectory is slowly climbing back toward the historical average. That is exactly the kind of cycle a bank-like company such as American Express has to absorb directly on its own balance sheet. Visa and Mastercard, in the same scenario, would feel only one thing: slightly fewer transactions flowing through their network. Never a single cent of loss on a loan they never made in the first place.

What this changes in my numbers, today

This mechanism is not just an accounting curiosity: it shows up plainly in the numbers I checked live on my site, on August 11, 2026.

Stock	Model	Net margin	Net debt / annual cash
Visa (V)	Open loop, never lends	50.8%	0.6 year
Mastercard (MA)	Open loop, never lends	46.3%	0.8 year
American Express (AXP)	Closed loop, lends directly	14.1%	12.5 years

That last column, net debt divided by the cash the company generates each year (in other words, how many years of cash it would take to wipe out all its debt), tells the whole story on its own. Visa would need a little over six months of its current cash to erase its net debt, Mastercard just under a year. American Express would need more than twelve years. This is not because American Express manages its debt poorly: it is because a large chunk of what shows up on its liabilities side is the loans it made to its own customers, which it has to fund upfront, well before collecting a single repayment. Visa and Mastercard simply have nothing comparable to fund, since they never lend a dollar to anyone.

A model is not automatically the better investment

It would be tempting to conclude that Visa's model is simply superior: less risk, fatter margin, end of story. That is exactly the trap I try to avoid with my method, which always separates the quality of a business from its price in the market. On business model quality alone, yes, never carrying credit risk is structurally more comfortable than carrying it. But American Express's closed loop also has an edge Visa lacks: it sees BOTH sides of every transaction (the customer AND the merchant), which lets it charge noticeably higher fees to the merchants that accept its card, and target a wealthier customer base with premium, higher-fee cards.

That is the whole point of the Membership Rewards program and of annual fees running into the hundreds of dollars: American Express is not just selling a way to pay, it is selling access to a customer base that spends more and pays its bills better than average, which is why a luxury hotel or an airline agrees to hand over a bigger slice of each transaction than it would concede to Visa. An open-loop network like Visa simply cannot sell that argument: it never knows the final customer's identity, only the flow between two banks that are strangers to each other. Checked live on August 11, 2026: Visa trades 26.0% above the reasonable buy price my model calculates, and American Express 67.1% above its own, while Mastercard, the third player in the same industry, actually trades 35.6% below its own. Three companies in the same industry, two different models, and no automatic conclusion to draw on which one to buy: model quality and the price you pay for it stay two separate questions.

Each model also carries its own political risk. Visa and Mastercard's open loop lives under the constant threat of regulatory caps on its interchange fees: debit card fees have already been capped by US law since the 2010 Durbin amendment, as this US Congressional Research Service report explains, and the same political pressure regularly resurfaces around credit cards. American Express, with its higher-fee model justified by a premium customer base, remains so far less exposed to that kind of cap, but carries in exchange a risk no regulatory ceiling can erase: the credit cycle itself.

The right reflex when you see these three stocks

Next time you see Visa, Mastercard and American Express listed side by side in an article or a stock screener, remember this: they are not three variants of the same company. They are two different businesses borrowing the same piece of plastic to exist in the public's eye. I already broke down American Express's quarter in depth in my analysis of its second quarter 2026 results, and Mastercard also shows up in my study of the one quality criterion that trips up even the largest stocks in the world. If you want to check where these three names stand today, my Visa, Mastercard and American Express pages run on the same live numbers quoted in this article.

FAQ

What is an "open-loop" payment network?

A network where a payment moves through four distinct players: the customer, their bank (which lends if needed), the merchant's bank, and the network (Visa, Mastercard), which simply moves the payment order between the two banks for a small fee. The network itself never lends any money.

What is a "closed-loop" network?

A network where a single company combines the roles of issuing bank and network, like American Express. It lends directly to its own customers and therefore carries the risk that a loan is never repaid, something an open-loop network like Visa never does.

Why can American Express charge merchants more?

Because it sees both sides of the transaction (customer and merchant) and targets a wealthier customer base, it can justify higher fees than Visa or Mastercard to the merchants that accept its card, in exchange for reputedly stronger customer spending power.

Does this mean Visa is a better investment than American Express?

No, not automatically. Never carrying credit risk is structurally more comfortable, but my method always judges a business's quality separately from its price in the market. As of August 11, 2026, Visa actually trades above my own calculated reasonable buy price, just like American Express, while Mastercard trades below its own.

About the author

Written by Lubin Danilo, founder of Lubin Investment. A self-taught individual investor, I find fundamental analysis fascinating, and it has delivered excellent results. For three years now, my performance has beaten the S&P 500. But analyzing every stock took too much time: sites with incomplete data, calculation methods and criteria never aligned with mine. And spotting the best stocks was just as time-consuming, even with my own well-defined checklist. So I put my software development background to work to build this software, base my investment strategy on its results, and share it with people who share the same passion as me. It judges a company's quality and its price separately, using criteria drawn from the financial literature (Warren Buffett, Michael Mauboussin, Aswath Damodaran).