Amex Charges Merchants More So It Can Pay Cardholders More. The Loop Is the Whole Trick.

A shopkeeper runs three cards for the same $100 sale and watches three slightly different amounts land in the account. Visa and Mastercard take their cut and hand most of it to the bank that issued the card. American Express takes a thicker slice — and keeps almost all of it. The shopkeeper grumbles about the third one and accepts it anyway. That last fact is the whole story: the merchant who hates the fee swipes the card regardless, because the person holding it spends more than the one holding the others.¹ Amex didn't win an argument about pricing. It built a machine where the argument never has to be won.

The official story is that Amex charges more because it's a premium brand serving rich customers, or because it has the market power to gouge. Both are tidy and both miss the mechanism. The fee is not a markup on the same product Visa sells. It is the output of an entirely different architecture — and once you see the architecture, the price stops looking like greed and starts looking like arithmetic.

Amex sits in all three seats Visa splits with banks

In a normal card transaction, three parties take economics: the bank that issued the card, the bank that signed up the merchant, and the network that connects them. On Visa and Mastercard those are usually three different companies — open-loop networks that rely on third-party banks to do the issuing and the acquiring. Amex runs a closed loop. It issues the card, it acquires the merchant, and it is the network in between, controlling the transaction end to end.⁶ So when a swipe happens, Amex doesn't take a thin network sliver and hand the rest to a partner bank. It captures the issuer economics, the acquirer economics, and the network economics in one pass. The higher merchant fee isn't a different price for the same service. It's what happens when one company collects three companies' worth of fees and answers to no partner about how it's split.

The fee funds the rewards that justify the fee

Here is where the loop closes on itself. Amex describes its own model as 'spend-centric' — it makes money primarily by driving card spending, and it says its cards spend higher on average per card than its network competitors do.¹ That extra economics it captured by owning all three seats doesn't just sit as profit. A large share gets recycled into rewards, lounges, and benefits that make the card worth carrying — and a card worth carrying gets pulled out more often, for bigger purchases. The merchant fee funds the rewards; the rewards drive the spend; the spend is exactly what the merchant is paying the fee to reach. It is a self-fulfilling toll. Merchants pay more because cardholders spend more, and cardholders spend more because merchants accept the card.

And the premium cardholder fee — the annual charge that makes people flinch — is the other half of the same engine. It pre-pays for the rewards before the spending even starts, and it self-selects the cardholders most likely to spend. Amex is not the only beneficiary; the volume it can promise a merchant is the thing that buys back acceptance. The clearest example is the Delta cobrand portfolio, which accounted for roughly 12% of Amex's worldwide billed business and about 21% of its card-member loans at the end of 2024.⁴ That is a single partnership delivering one-eighth of all spend on the network — the kind of concentrated, high-value volume that makes a merchant think twice before turning the card away.

But isn't the fat fee just gouging that's finally being competed away?

The honest objection is that this whole story romanticizes simple market power — Amex charges more because it can, and the premium is eroding precisely because the power is fading. There's truth in it. The often-repeated 'Amex costs 3%, Visa costs 2%' framing is now stale for most small merchants. Through OptBlue, third-party processors contract directly with small merchants and set their own pricing on Amex's network — not Amex — and rates for small merchants have moved toward those of Visa and Mastercard at the low end.⁵ Amex's own filing admits its merchant rates are under downward pressure from competitor pricing.³ So the toll is genuinely being competed down where it was always softest. But notice what that concession actually proves: the spread survives precisely where Amex can still promise a merchant volume worth paying for — large, directly negotiated relationships — and erodes where it can't. The fee was never a fixed monopoly rent. It was always the price of access to a particular pool of spend, and it floats with how much that access is worth.

The closed loop is also no longer purely closed, which complicates the neat story in Amex's favor. Amex now runs revolving credit at scale, not just charge cards, and lets third-party banks issue on its network too. The 'just a charge-card company that rejects merchants' picture is a museum piece: in the U.S., Amex reports acceptance at 99% of places that take credit cards — what it calls "virtual parity" with Visa and Mastercard — and has grown its global merchant footprint by nearly 5x since 2017.⁹ The acceptance argument against Amex was real twenty years ago. It is now mostly an international footnote.

Amex's higher fee was never charity for merchants, and it was never brute pricing power either. It is the visible output of an invisible decision — to own every seat at the table instead of renting one — and then to spend the proceeds buying back the spending that makes the seat worth owning. The merchant pays more because the cardholder spends more; the cardholder spends more because the merchant pays more. Cut the loop anywhere and the whole thing slackens. That's why the premium is being competed away at the edges and holding firm at the core: the toll is only ever worth what the road it guards is worth — and Amex spent a century making sure it owned the road.