Amex's Moat Isn't Rich Customers. It's a Legal Shield and a Closed Loop.

Pull out a green Amex and the story tells itself: this is the card the wealthy carry, accepted where money goes, declined where it doesn't, protected by a customer base too well-off to default. In 2024 the company posted a record $10.13 billion in net income on $65.95 billion in revenue, with net card-fee revenue up 16% to $8.4 billion.² The affluence narrative looks airtight. It is also pointed at the wrong wall.

The official story is that Amex's moat is its customers' bank accounts. The real moat is structural and legal — a closed transaction loop and a set of merchant rules a sharply divided Supreme Court declined to strike down. The income story is a consequence of that machinery, not the source of it. And the machinery is being quietly compressed from two directions at once.

The moat is the loop, not the bank balance

Visa and Mastercard run open, four-party networks: a card-issuing bank, a merchant-acquiring bank, and the network wire in between. The data and the rewards are split across parties. Amex, for most of its proprietary cards, is its own issuer, its own network, and its own acquirer — a closed loop. That single architectural choice is the moat's floor. Because Amex sits on both ends of its own transactions, it sees the full spend, sets the rewards, and prices the merchant directly. On $1.551 trillion of billed business across 83.6 million proprietary cards in 2024¹, that closed loop is one of the richest first-party spending datasets in commerce. A rival can copy a metal card and an airport lounge. It cannot copy two decades of seeing exactly what its cardholders buy, where, and when — without rebuilding the loop from scratch.

The customer-quality story is real, just secondary. Amex's Q4 2024 net write-off rate ran around 1.9%, against roughly 4.17% for commercial-bank credit cards generally.⁸ That is a genuine edge — lower losses mean richer rewards funded out of the same fee. But it is the loop that lets Amex capture and monetize that quality. The affluence is the cargo. The loop is the road.

The wall around the loop is a 5-4 ruling

Here is the part the income narrative misses entirely. Amex's higher merchant fee only survives because merchants are forbidden to talk customers out of using the card — no 'we prefer Visa,' no surcharge at the register. Those anti-steering rules are what let Amex hold price. In 2010 the Department of Justice sued Amex, Visa, and Mastercard over exactly those provisions. Visa and Mastercard settled. Amex fought it to the Supreme Court — and in 2018 it won, but by a single vote.⁵ The Court held that in a two-sided transaction market, competitive harm has to be weighed on both sides at once, and on that framing Amex's rules survived Section 1 of the Sherman Act.⁵

Read that holding carefully. It resolved one antitrust framing question — how to define the market — by the narrowest possible margin. It did not immunize the rules forever. And Amex itself behaves as if it knows that: its 10-Q acknowledges ongoing downward pressure on merchant discount rates from regulatory caps, and the company has already exited its network business in the EU and Australia under interchange regulation.⁶ The legal shield that holds the price is real. It is also a single statute, a single vote, and a regulatory regime that has already pushed Amex out of two developed markets.

The ceiling: one airline, and a fee gap that closed

If the loop and the legal shield are the moat's floor, two things are pressing on its ceiling. The first is concentration. The 'diversified affluent base' story implies spending spread thinly across millions of independent wealthy households. Amex's own 2024 10-K tells a more concentrated story: the Delta cobrand portfolio alone was roughly 12% of worldwide billed business and roughly 21% of worldwide Card Member loans, with the agreement running through the end of 2029.³ One airline relationship anchors more than a tenth of the spend and a fifth of the lending book. That is leverage Delta knows it holds at every renewal.

The second pressure is the price itself. The folklore says Amex charges merchants far more than the open networks, and that premium funds the rewards that attract the affluent. But under the OptBlue program, which brought millions of smaller merchants onto Amex acceptance, wholesale rates can run as low as 1.6% plus $0.10 per transaction depending on industry and volume — directly comparable to Visa and Mastercard interchange in many categories.⁸ The wide spread that the moat narrative leans on has materially narrowed. Amex traded fee-per-swipe for acceptance breadth, which is rational, but it means the premium that was supposed to be the moat is no longer the moat.

And note the structural cost the popular story underweights: unlike Visa and Mastercard, which push credit risk onto issuing banks, Amex carries its own loans. The low write-off rate is a strength right up until a downturn hits upper-income consumers — at which point the affluence that looked like a defense becomes direct exposure on Amex's own balance sheet. Amex says as much itself, warning that premium consumer spending is sensitive to discretionary levels and falls in downturns.⁴

But the numbers are at record highs — doesn't that settle it?

The fair objection is that this reads too bearishly against a company printing records: $10.13 billion of net income, 13 million new cards acquired in 2024, card-fee revenue up 16%.² If the moat were compressing, why are the fees accelerating? The honest answer is that the moat is real and still working — the question is what's actually doing the work. Discount revenue, the merchant-fee line, has made up more than half of Amex's revenue in recent quarters⁷, and that is exactly the line OptBlue convergence and regulatory caps press on. Meanwhile small-business and corporate clients made up about 42% of billed business in 2024⁴ — diversification away from the consumer-affluence story the brand sells. The records are genuine. They are also being produced by a machine whose two strongest walls — the price premium and the legal shield — are the two most exposed to erosion. A moat can be wide and narrowing at the same time.

Amex's moat was never the size of its customers' bank accounts. It was the decision, made a century before anyone said 'two-sided market,' to own the whole loop and then to defend the rules that let it price the loop. That moat is still standing, and still profitable. But the floor is a 5-4 ruling and the ceiling is a single airline and a fee that has quietly converged on its rivals'. The genius is in the architecture. So is the fragility — and the affluence story is built to look at neither.