Visa Pays One Side and Bills the Other. That Asymmetry Is the Flywheel.

Notice who is happy at the checkout counter and who is not. You, the cardholder, are quietly subsidized — points, cashback, fraud protection, the ability to spend money you don't yet have. The merchant, meanwhile, is grimacing, because every swipe costs them a cut they can't refuse. That asymmetry isn't an accident or an oversight. It is the engine. Visa figured out long ago that the way to spin a payments network up to enormous scale is to pay one side to show up and bill the other side for the privilege of meeting them.

The official story is that Visa is a credit-card company that charges everyone a fee. It is wrong twice over. Visa doesn't lend, and it doesn't even collect the fee most people think it lives on. What it actually runs is a two-sided market with a deliberately lopsided price — and that lopsidedness is what makes the whole thing compound.

Subsidize one side, tax the other, and watch it spin

Strip the network down to its mechanics and you find two crowds Visa has to keep growing at once: cardholders and merchants. Each is worthless without the other. A card no shop takes is a coaster; a shop with no cardholders walking in is just expensive plumbing. So the network has a chicken-and-egg problem at birth — and the way Visa solved it was to make one side of the egg cheap. Cardholders are courted with rewards and protection that often exceed what the card costs them; merchants are charged for access to the spenders those rewards attract. The Supreme Court actually blessed this logic in Ohio v. American Express: a two-sided payment platform can rationally price one side below cost to subsidize the other, because it isn't selling to two markets — it's selling one connection that both sides need.⁸ The subsidy isn't generosity. It's bait, and the bill goes to the other table.

Here is the part nearly everyone gets backwards. The merchant's pain — the interchange — does not land in Visa's pocket. Visa's own published interchange document is blunt about it: 'Merchants do not pay interchange reimbursement fees; merchants pay merchant discount to their financial institution.'⁵ Interchange flows from the acquiring bank to the issuing bank, between financial institutions, never through Visa's income statement. Visa told the SEC the same thing directly: its fees 'are not derived from interchange reimbursement fees or merchant discount rates.'⁴ So Visa sets the schedule that splits the spoils, then charges both banks a separate toll for using its rails. It designs the casino's payout tables and takes a fee on every chip moved — without ever betting.

Why each turn of the wheel makes the next one easier

A flywheel is only a flywheel if the output feeds the input. Watch the loop. Subsidized cardholders make a Visa credential worth carrying, so issuers hand out more of them — 4.6 billion are now in wallets and phones.² More credentials make accepting Visa unavoidable for any serious merchant, so acceptance spreads past 150 million locations.² Wider acceptance makes the credential even more useful, which pulls in still more cardholders, which makes acceptance even more mandatory. Round and round. In fiscal 2024 the wheel was turning $16 trillion in total volume across roughly 303 billion Visa-branded transactions.² And the punchline is what that volume costs Visa to handle: almost nothing extra per swipe. The rails were built once; each new transaction is nearly free to process but still carries a fee. So $16 trillion of motion compounds into $35.9 billion of net revenue, up 10% in a single year.¹ The flywheel doesn't just spin — it spins faster the bigger it gets.

The wheel was built to survive its own owners

The durability was baked in early, almost by accident of structure. Bank of America launched BankAmericard in 1958, but the thing we now call Visa didn't exist until 1970, when Dee Hock organized the rival issuing banks into an independent consortium — National BankAmericard Inc. — owned by the very banks that competed using it.⁷ The brand became Visa in 1976. That ownership design matters more than the name: the network's customers were its members, which meant every bank had a stake in feeding the flywheel rather than starving it. By the time Visa went public in 2008 — selling shares at $44 to raise roughly $17.9 billion in the base offering, near $19.1 billion with the overallotment, the largest U.S. IPO of its day — the loop was already self-sustaining, no longer dependent on any one bank's enthusiasm.⁶ A network that needs nobody in particular is very hard to kill.

Isn't this just a tax regulators will eventually cap?

The fair objection writes itself: this is rent on commerce, the subsidized side is bait, and the taxed side has been complaining loudly enough that surely the law catches up. It's a real threat, and it isn't theoretical — interchange has been litigated and legislated against for years. But notice what the asymmetry does to its own defenders. Because Visa doesn't collect interchange, the fee everyone wants to attack isn't even Visa's revenue line; cap it, and you've reshuffled money between two banks while Visa's network fees keep flowing on the same volume.⁴ And the deeper problem for would-be regulators is that the courts have accepted the economics: subsidizing one side to grow the platform is rational, not predatory, per Ohio v. American Express.⁸ The honest counter is that the rent is genuinely under pressure at the edges — bank-to-bank instant rails can route around a card network entirely where they take hold. But the flywheel's defense was never the fee. It was the two crowds, each holding the other in place, that no single law and no single rival can pry apart at once.

Visa didn't win by being the cheapest, the cleverest, or even the lender. It won by standing between two crowds and pricing them in opposite directions — paying the spenders to arrive and billing the sellers who needed them. Every swipe deepens the asymmetry that made the swipe possible. The flywheel runs on its own lopsidedness, and the longer it spins, the more the world organizes itself around the assumption that it always will.