What it does · how it got here · why it's not Visa
If you have read our note on Visa, you already understand half of American Express — and the other half is precisely what makes it different. Visa is a toll road: it runs the network for other people's cards and clips a sliver, touching no customer and lending no money. American Express is something rarer — a closed-loop franchise that does the whole job itself: it issues the card, runs its own network, signs up the merchants, and lends to the cardholder. It sees both ends of every swipe — who spent, and where — and it makes money three ways at once. Where Visa is a pure toll collector, Amex is a toll collector, a bank, and a luxury club rolled into one green-and-gold card.
Founded in 1850 — as a freight-and-express company, before it ever touched money — Amex became, over 175 years, the premium name in payments. It targets not everyone, but the affluent, high-spending customer, and charges merchants a richer fee to reach them. In 2025 it earned about $80 billion of revenue and $10.8 billion of profit, at a return on equity of ~34%. The stock trades near $348, worth about $240 billion — and its largest shareholder, holding roughly a fifth of the whole company and never having sold a share in thirty years, is a man named Warren Buffett.
"You can't create another American Express. You can't create another Coca-Cola. The value is in the minds of people — and that's a wonderful moat." — Buffett, on why Amex has been in Berkshire since the 1960s
That is the whole thesis in a sentence. Amex is not the biggest network — Visa and Mastercard dwarf it in transactions — but it owns something they do not: a premium brand and a direct relationship with the world's best-spending customers. That is a different kind of moat from Visa's, aimed at a different kind of customer, and — as we'll see — priced today at a distinctly friendlier multiple. It also carries a risk Visa never touches: because Amex lends its own money, a recession bites it in a way it can never bite a toll road.
| Founded | 1850 · New York (as a freight company) |
| Sector / Industry | Financial Services · Credit-Services network |
| Model | Closed-loop — issuer + network + acquirer + lender |
| The customer | Affluent, high-spending consumers & businesses |
| Revenue (FY2025) | ~$80 B (blends fees, card fees & net interest) |
| Market capitalisation | ~$240 B |
Closed-loop vs. open network · the three ways Amex gets paid
To own Amex honestly you must see how its "closed loop" differs from Visa's "open network," because it changes both the reward and the risk. Here is the whole machine:
Amex gets paid three ways, and the mix is the key to the business:
1. Discount revenue (the biggest). The fee merchants pay Amex on every transaction — and it is higher than Visa's or Mastercard's, because Amex delivers premium, high-spending customers. The pitch to the merchant is simple: "we bring you the best customers, so you pay us a little more." That is pricing power earned by the quality of the cardholder.
2. Card fees (the fastest-growing). The annual fees on premium cards — the $695 Platinum, the Gold, the business cards. This is a beautiful, recurring, high-margin stream that also locks the customer in (having paid for the club, you use it), and it now runs to billions a year, growing at a double-digit clip.
3. Net interest income (the bank part). Because Amex lends its own money on revolving balances, it earns interest — about $17 billion of it. This is the engine that doesn't exist at Visa, and it is a double-edged sword: extra profit in good times, credit losses in bad ones.
The segments · a US-centric footprint · the size of the pond
Amex reports across three businesses, all serving the same premium franchise. [FMP's segment feed mixes old and new labels; the split below is approximate.]
Here is the strategic contrast with Visa in one line: Visa is broad and global; Amex is premium and American. Visa earns a sliver from nearly every card swipe on earth; Amex earns a fat fee from a chosen slice — the big spenders — mostly in the United States. That focus is both its moat (nobody owns the affluent relationship like Amex) and its limit (it will never be as universal as Visa). The growth engine is a happy one, though: a relentless refresh of premium cards, a successful courtship of younger affluent customers (Millennials and Gen-Z now drive most of the new-card growth), and a long international runway.
Brand, the affluent closed loop, and the rewards flywheel
Amex's moat is real and durable, but it is a different moat from Visa's — narrower, deeper, and built on prestige rather than ubiquity. It is fed from three springs.
The brand. American Express is one of the great aspirational brands in finance — the card that signals you have arrived. That prestige lets Amex charge annual fees a commodity card never could, and lets it charge merchants more. This is the intangible-asset moat Buffett prizes, the same kind that lets Coca-Cola charge up for sugar water — "the value is in the minds of people."
The closed loop. Because Amex owns both ends — cardholder and merchant — it sees every transaction end to end. That gives it data, direct relationships, and a control over the experience (rewards, fraud, service) that an open network splitting the job across thousands of banks simply cannot match.
The rewards flywheel. Fat merchant fees and annual fees fund rich rewards (Membership Rewards, airport lounges, hotel and airline partners); rich rewards attract affluent big-spenders; big-spenders make the network more valuable to merchants, who pay the fat fees. Round it goes.
How durable? Wide and stable — for its niche. Within the premium segment, this moat is formidable and, if anything, widening as Amex wins the next affluent generation. The honest limits: it is not the universal toll road Visa is (fewer merchants accept it, though the gap keeps closing), and the same fintechs and real-time rails that nibble at Visa nibble here too. But the affluent-brand franchise is among the most defensible in finance.
Squeri at the wheel · and the most famous shareholder in the world
On capital and stewardship, Amex has one of the finest endorsements money cannot buy — but let's start with the people running it day to day.
American Express is not just a Berkshire holding; it is one of the founding legends of Buffett's career. In the 1960s, when the "Salad Oil Scandal" briefly panicked the market into thinking Amex might be ruined, a young Buffett noticed that diners were still happily paying with their green cards — the brand was untouched — and made one of his first great concentrated bets. He returned in the 1990s, and Berkshire has held roughly a fifth of the company ever since, never selling a single share.
Why does this matter to you, beyond the glamour of the name? Because it is a live demonstration of the single most powerful idea in this whole series: a wonderful business that retires its own shares turns a patient owner richer every year, doing nothing. Buffett has taken his stake from a eighth to a fifth of Amex without lifting a finger, simply because management keeps buying back stock. That is the compounding machine you are buying into — provided, as always, you buy it at a sensible price.
Sourced from your endpoints · a lender, so read it differently · linked to the data pages
| Metric | Value | Read |
|---|---|---|
| Revenue growth (5-yr) | ~13% CAGR | ▲ double-digit |
| Net income (FY2016 → FY2025) | $5.4B → $10.8B | ▲ ~2× · EPS ~2.7× |
| Return on equity (ROE) | ~34% | ▲ exceptional for finance |
| Net profit margin | ~13.5% | ◆ read vs a lender, not a toll road |
| Net interest income | ~$17B | ◆ the lending engine — & the risk |
| Free cash flow / yield | ~$14B · ~6% | ▲ far richer yield than Visa |
| Dividend / payout | ~1.1% · ~21% | ▲ well-covered, growing |
| Book value / P/B | $49.6 · ~7.0x | ◆ high P/B, but ROE justifies |
| Capital / stress tests | passes · min buffer | ▲ well-capitalised |
| Altman-Z | 0.7 | ◆ ignore — meaningless for a lender |
A word on how to read these numbers, because Amex is part-lender and the usual staple-company yardsticks mislead. Its ~13.5% net margin looks thin next to Visa's ~50% — but that is because Amex's revenue includes the interest it earns and pays as a bank; on the measure that matters for a lender, return on equity of ~34%, it is world-class. Ignore the Altman-Z score of 0.7 entirely — it is designed for manufacturers, and flags every healthy lender as "distressed." The two numbers to actually watch are credit quality (Amex's charge-offs run well below the industry because its customers are affluent) and capital (it comfortably passes the Fed's stress tests). This is a financially strong franchise — but it is a lender, which is why I score its balance-sheet strength a notch below the toll-road purity of Visa.
Cheaper than Visa · DCF vs. a lukewarm Street · margin of safety
| Yardstick | Amex today | Visa, for contrast | Read |
|---|---|---|---|
| P/E — reported earnings | ~21x | ~30x | materially cheaper |
| Free cash flow yield | ~6% | ~3% | twice the yield |
| PEG (P/E ÷ growth) | ~1.5 | ~1.9 | fairer for the growth |
| Dividend yield | ~1.1% | ~0.8% | a grower, plus buyback |
| Return on equity | ~34% | ~50%+ | both superb |
| Credit risk | yes (it lends) | none | the price of the cheaper multiple |
Here is the case in plain terms. American Express is a wonderful franchise growing its revenue at double digits and its earnings faster still — and it trades at about twenty-one times earnings, a full third cheaper than Visa's thirty, with twice the free-cash-flow yield. On the face of it, that is the more attractive of the two payment franchises. The catch, and it is a real one, is the source of the discount: Amex lends, so a recession that would barely dent Visa's toll would raise Amex's credit losses and dent its earnings. That cyclicality is exactly why the market pays less — and why a lukewarm Wall Street sits at a mere "Hold" despite the quality. The mechanical DCF (~$220) is too bearish to trust much on a lender; the truer read is a good business at a fair-to-slightly-cheap price, with the recession risk the thing you are being paid a lower multiple to bear.
Credit cycle · acceptance · competition · regulation — with sources
Amex's defining risk — the one that separates it from Visa — is the credit cycle: because it lends its own money, a sharp recession raises charge-offs and dents earnings in a way a pure network never suffers. Its affluent base makes those losses far milder than a subprime lender's, but they are real, and they are why the stock is cyclical. Around that sit the ordinary worries: a merchant-acceptance gap versus Visa and Mastercard (narrowing, but still there), fintech and real-time rails that could disintermediate cards over time, a heavy US concentration, and an ever-costlier rewards arms race to keep the affluent loyal. None threatens the franchise. Together they are why a business this good trades at only ~21× — and why you are paid, in a lower multiple, to accept the cyclicality.
I have owned this company, in one form or another, for the better part of sixty years — and I want to explain why it has earned a permanence in our affairs that almost nothing else has, and why, even so, I would have you mind the price.
American Express is a wonderful business, and it is wonderful in a way I understood long before I understood much about technology. When I first bought it, in the 1960s, a scandal had frightened the market into thinking the company might be finished — and I went and sat in a Omaha steakhouse and watched people pay, cheerfully, with their green cards, exactly as they always had. The brand was untouched. That is the whole secret of this business: the value sits in the minds of people — the affluent customer who feels he has arrived when he pulls out the card, and the merchant who pays a richer fee because that customer spends more than anyone else. You cannot build that in a laboratory or buy it with a billion dollars. It is a moat made of reputation, and it is one of the most durable kinds there is.
Notice, too, how different it is from Visa, which we admired in these pages but found dear. Visa is a magnificent toll road that touches no customer and lends no money; Amex does the whole job itself — issues the card, runs the network, courts the merchant, and lends its own money besides. That last part is the crucial difference for an owner: because Amex lends, it earns more per customer than a toll road ever could, but it also carries a credit risk that a recession will surely, one day, test. I do not say that to frighten you — Amex's customers are the best in the business and its losses run well below the industry — but honesty requires it. This is a wonderful franchise with a bank attached, and the bank is why the market, quite reasonably, pays a lower multiple.
And what a multiple. Here is the part I find most agreeable: this fine business, growing its revenue at double digits and its per-share earnings faster still, changes hands at about twenty-one times earnings — a full third below Visa, with twice the cash yield. Wall Street, fretting about the next recession, can only bring itself to say "hold." I have never let Wall Street's mood set my own. When a business this good is offered at a fair price, and its own management keeps quietly buying back its shares — lifting our ownership from an eighth to a fifth of the company without our spending a dime — a patient owner does very well simply by sitting still and letting the machine work.
So here is my counsel. If you own American Express, hold it, and let the buyback and the growing dividend compound your slice — this is a business to keep for a very long time. With fresh money, I would begin accumulating here, at a fair price for a franchise this fine, mindful that it is more cyclical than a toll road and sizing it accordingly. And I would keep some powder dry for the day a recession scare drags it toward the high $280s — because the surest time to buy a wonderful lender is precisely when the crowd is most afraid of the credit cycle, and the price finally reflects that fear. Be greedy, as I have said, when others are fearful. Amex has rewarded that patience for sixty years, and I see no reason it will not reward it for sixty more.