Everything earlier in this module described a model that looks ripe for disruption. Interchange is a tax merchants resent, scheme fees keep climbing, and least-cost routing exists precisely because the rails are not the cheapest way to move money. So a fair question is why Visa and Mastercard still sit in the middle of nearly every card transaction. The honest answer is that price was never the thing holding the system together. Distribution, defaults, and the math of two-sided networks are.
If you build payments, you need to understand these forces concretely, because they are the same forces that will either carry your product or quietly starve it.
Start with the numbers
Visa and Mastercard together cleared roughly $9.99 trillion in US purchase volume in 2025. Of that combined total, Visa carried about 70 percent and Mastercard about 30 percent, with American Express and Discover making up the remainder of the broader market. That concentration is not an accident of branding. It is the visible output of the mechanisms below.
Keep one distinction in mind from the four-party lesson. Visa and Mastercard do not issue cards or sign merchants directly. They license that work to issuers and acquirers. Their durability comes from sitting at the one point both sides already agreed to plug into.
The two-sided cold-start problem
A payment network is worthless to a cardholder if no merchant accepts it, and worthless to a merchant if no cardholder carries it. Both sides have to show up at once, and neither has a reason to be first. This is the cold-start problem, and it is the single hardest thing about launching a new rail.
The practitioner rule, well captured in CGAP's work on two-sided merchant payments, is that you build out the harder side first. For card-style schemes that side is acceptance. If you excite consumers before acceptance exists, they hit dead ends, churn, and the product rarely recovers.
Why this protects incumbents specifically
Visa and Mastercard already solved the cold start decades ago, so a new entrant is not competing on a feature. It is being asked to rebuild acceptance at merchants who already accept the incumbents and feel no pain that a second network fixes. The incumbent's installed base is the moat, and it compounds. Every new merchant terminal and every issued card raises the value of the network to the other side, which is why acceptance footprint, not transaction price, is the asset that matters.
History rhymes here. Diners Club, conceived in 1949 and launched in 1950, brought out the first general-purpose charge card by charging merchants 7 percent, a number only tolerable because the card brought in free-spending customers merchants could not otherwise reach. The network was bootstrapped on a value proposition strong enough to overcome the cold start. Absent that, you get nowhere.
Defaults and distribution
Most payment behavior is not chosen at the moment of payment. It is set far upstream, by what the issuing bank puts in the customer's hand and what the checkout flow defaults to.
When a bank issues a card, it picks the network as a business decision, then ships millions of cards on that rail. The cardholder almost never thinks about it. On the merchant side, the acquirer and the gateway decide which rails are wired in. By the time a transaction happens, the network was effectively chosen weeks or months earlier by two intermediaries, neither of which is the end user. An entrant has to win those intermediaries, not the consumer, and the intermediaries optimize for the rail their customers already expect.
This is why distribution beats product in payments more often than in almost any other category. The incumbent is the default in every flow that already exists.
Switching costs, and where they actually live
Switching costs in this market are concentrated at the portfolio level, not the individual cardholder level. A single consumer can get a new card in minutes. Moving a large co-brand book is a different exercise entirely.
Worked example: the Costco conversion
In 2015, Costco announced it would drop American Express and move its US co-brand to Visa with Citi as issuer. The switch took effect on June 20, 2016. Citi acquired the roughly $10.5 billion American Express card portfolio, and more than 11 million cardholders had to be reissued and migrated.
The economics that justified that effort were large. BMO Capital Markets estimated Costco could save on the order of $110 million to $220 million a year on acceptance costs, with typical Amex acceptance around 190 to 200 basis points versus an expected 130 to 140 for Visa and Citi. Even with savings that size on the table, the conversion was a multi-year, heavily negotiated, operationally heavy project involving reissuance, systems work, and a portfolio sale.
That is the real shape of switching costs. They are not a UX inconvenience. They are reissuance logistics, integration work, contract terms, and the risk of disrupting customers mid-relationship. The friction is what makes incumbent positions sticky even when a cheaper rail exists, and it is why a merchant or bank needs a payoff in the hundreds of millions before it moves a major book.
Where the model actually bends
The mechanisms above are strong, but they are not absolute, and you should not treat the incumbents as permanent. The clearest counterexamples come from account-to-account rails that sidestep the four-party model rather than competing inside it.
Brazil's Pix, run by the central bank, reached about 70 percent of the population and roughly 15.4 billion transactions in the second quarter of 2024. India's UPI reached around 350 million users and a large share of global real-time transactions. Both beat the cold start the same way: a central operator mandated interoperability across banks and fintechs, so acceptance and reach arrived together rather than one side waiting on the other.
The lesson for builders is precise. The incumbents are hard to dislodge inside their own model, where defaults, distribution, and acceptance footprint all favor them. They are most vulnerable to a different model that solves the two-sided problem by fiat or by a value proposition strong enough to move both sides at once, which is exactly what the next lesson on pressure points takes up.
Takeaway
Incumbent card networks survive on installed base, not on price or technology. The moat is the acceptance footprint built by solving the cold start long ago, reinforced by defaults set upstream by issuers and acquirers, and locked in by portfolio-level switching costs that demand large, certain payoffs before anyone moves. If you are building a rail, do not plan to out-feature the incumbents inside their model. Plan to solve the two-sided problem on both sides at once, or find the path that routes around it.