Everyone Says Amex's Moat Is the Closed Loop. The Closed Loop Just Got Cloned.

For decades, the standard answer to the question "why can't anyone copy American Express?" came in two words: closed loop. Amex issues its own cards, owns the cardholder relationship, processes the transaction, and negotiates the merchant fee — all under one roof, capturing the full economics that Visa and Mastercard hand off to a chain of banks.⁸ It was the textbook moat. Then, in February 2025, Capital One paid $35.3 billion for Discover and bought itself a closed loop of its very own — a network it could route its own cards through, internalizing the same advantage.⁶ The unique structural edge turned out to be for sale. So the question is no longer rhetorical. With the loop now cloneable, what actually protects American Express?

The official story is that the closed loop is the moat. The truer story is that the closed loop is the plumbing — and the moat is the thing flowing through it. Amex's real defense isn't a corporate structure anyone with $35 billion can buy. It's a flywheel that took a century to spin up and would take a competitor far longer than a single acquisition to match.

The flywheel that the loop only makes possible

Here is the loop running at full speed. Amex attracts affluent, high-spending cardholders — people who put real money through the card. In 2024 that meant $1,551 billion of billed business across just 83.6 million proprietary cards, a spend-per-card that no mass-market issuer comes close to.¹ Because those cardholders spend so much, merchants tolerate a higher fee to accept the card; they're paying to reach a customer who buys. That above-market merchant fee — discount revenue, Amex's largest line³ — funds rich rewards. The rewards attract the next batch of premium cardholders. And around it goes. The closed loop matters only because it lets Amex keep the full fee instead of sharing it with an issuing bank — the loop is the funnel; the spend is the water.

Capital One can now own the same funnel. What it cannot buy off a shelf is a hundred-year-old base of cardholders conditioned to put their largest purchases on the card, and a merchant base that has decided those customers are worth a premium to reach. You can buy a network in a quarter. You cannot buy the spending habit that makes the network valuable. That is the part the "closed loop is the moat" framing always missed.

And the flywheel still recruits its next generation. By Q2 2024, Millennials and Gen Z accounted for 33% of total billed business — the highest share that cohort has ever delivered — and 77% of new accounts that quarter were premium cards.⁷ The machine that converts young high-earners into lifelong heavy spenders is still turning. That's what a flywheel does: it doesn't just defend the present, it manufactures the future customer.

Where the water leaks out

A flywheel is only as strong as the fee that funds it, and that fee is quietly thinning. Amex's own 2024 10-K says discount revenue rose on volume but was partially offset by lower average merchant discount rates.³ The popular shorthand of a uniform 2.5–3.5% merchant fee is already a legacy artifact; the headline rate is in secular decline as Amex courts smaller merchants on cheaper terms to win acceptance. Each basis point shaved off the merchant fee is a basis point less to spend on rewards — and rewards are the centrifugal force keeping the wheel spinning. The threat isn't that a competitor copies the structure. It's that the economics inside the structure slowly compress.

The second leak is concentration, and it has a precedent. In 2015, when the Costco partnership ended, Costco was 8% of global billed business, 20% of worldwide loans, and 10% of cards-in-force — a single partner whose departure forced a painful reset, cushioned only by the fact that Costco paid a below-average discount rate.⁵ The market treated it as a near-death scare. Now look at the current portfolio: the Delta cobrand was approximately 12% of worldwide billed business and approximately 21% of card-member loans at the end of 2024.² That is a larger single-partner concentration than Costco ever was. The agreement runs through 2029² — but 2029 arrives, and the renegotiation will be conducted by an airline that knows exactly how much of Amex's wheel it can stop.

But isn't lending the part Visa was smart to avoid?

The fair objection cuts the other way: maybe the real vulnerability isn't concentration at all, but the fact that Amex lends. Visa and Mastercard famously bear no credit risk — they take a sliver of volume and let the banks eat defaults. Amex does the opposite: it carries the loans on its own balance sheet, which is why net interest income was $15.5 billion in 2024 and why it had to set aside $5.2 billion against credit losses that same year.¹ In a downturn, that provision line is the one that bites. So isn't the lending exposure the soft underbelly?

It's a real risk, but it's also the wrong frame, because the lending is downstream of the same flywheel. Amex lends to a curated base of high-spend, high-income cardholders — the very population the premium funnel selects for. The credit risk it carries is a better book than a mass-market lender's, precisely because the moat does the screening upstream. The $15.5 billion of interest income isn't a separate, riskier business bolted on; it's the flywheel earning a second revenue stream off the same customers. The danger isn't that Amex lends. It's that the two leaks — compressing discount rates and outsized partner concentration — could thin the rewards and shrink the spend at the same moment a credit cycle turns. The wheel survives any one of those. It's the convergence that would hurt.

American Express spent a century building something that looked, from the outside, like a clever corporate structure. In February 2025, a competitor proved the structure was for sale. But the structure was never the point. What protects Amex is the flywheel the structure was built to spin: premium spenders pulling in merchants, merchants funding rewards, rewards pulling in the next premium spenders — a loop a rival can replicate the architecture of and still spend a decade failing to fill. The moat was never the loop. It was the spend the loop was built to carry — and the only way to lose it is to let the fee that funds it thin, or to lean so hard on one partner that they learn how much of the wheel they hold.