Priceless: $MA Is Best in Class

Nobody is choosing a store because of which network their card runs on (besides those that are Costco customers, who can point to Visa). The majority have, at most, a vague feeling about their card; the points, the color, the annual fee that keeps you annoyed but you always forget to cancel until you’re charged; but no feeling whatsoever about the plumbing underneath it, which quietly takes roughly 0.27% of every dollar you have ever tapped, swiped, or autofilled. The best businesses ever built are the ones you never think about, and for almost two decades the market has understood this perfectly: it paid thirty-plus times earnings for the privilege of owning plumbing without the poop.

Then, over about the last year or so, the market has changed its mind. Mastercard ($MA) is down 13.2% in 2026. American Express ($AXP) is down 13.9%, the aforementioned Visa ($V) with its exclusive Costco deal is down 7.3%, the relative survivor in the sector, and at the June 3 trough Mastercard and Amex were both off more than 21% from their highs. The reasons come in threes: stablecoins are going to route around the networks, regulation from DC is threatening to cap the fees, and AI agents — this is the newest one — will supposedly shop on whatever rail saves them a nickel, because a robot doesn’t care about airport lounges.

Then these companies reported Q1 results. Mastercard grew revenue 16% and adjusted earnings 23%. Visa grew revenue 17% (its fastest quarter since 2022) and answered the panic with a fresh $20 billion buyback announcement. Amex grew earnings 18% and reaffirmed its full-year guide. Not one line of one income statement shows the disruption the drawdown is pricing. The fear is saying the model is dying, that there is disruption in the industry. The filings say that it’s compounding.

But “the card stocks are cheap” is a screen, not a thesis, because the three names underneath the panic are not the same animal. Visa and Mastercard are pure networks: tolls with software margins and not a dollar of credit risk. American Express is a network that is also a bank — it issues the cards, lends the money, and eats the losses — which means the same headline lands on three different income statements in three very different places. One fear, three exposures, three different prices being offered for them. This is a horse race. The winner is Mastercard, and I’m going to make you sit through the whole race to earn the verdict.

The Best Business Model Ever Invented

Visa ran a 50% net margin on $40 billion of fiscal 2025 revenue; half of every revenue dollar dropping to Generally Accepted Accounting Principles (GAAP) net income (Mastercard runs in the mid-forties). For context, my Nvidia piece spent the entire time marveling at a 56% net margin as the best in megacap land and Visa has been doing something within shouting distance of that every year for over a decade, without a product cycle, without inventory, without a factory to depend on, without a single dollar of credit risk. The networks don’t lend. They don’t carry receivables. When the economy sours, the banks eat the losses and the networks just process slightly fewer transactions. There is no other business at this scale with this margin and this little balance-sheet risk. There is honestly nothing else close.

And the machine has been astonishingly generous to anyone who ignored the recurring obituaries. Mastercard came public in May 2006 at $39 a share (I wish I was investing when I was 10!) — $3.90 adjusted for the 2014 ten-for-one split. At $495.24 as of the June 9, 2026 close (I’m publishing this on June 12th, so the price will have moved by the time you read it), that’s roughly a 127-fold return in twenty years, before dividends. Visa’s March 2008 IPO was the largest in American history at the time — they priced the biggest equity offering ever into the teeth of the financial crisis and it worked — and the split-adjusted $11 cost basis has returned roughly 30x. These two stocks spent fifteen years as the closest thing public markets had to a cheat code, which is exactly why a 13% drawdown on no earnings damage deserves attention and confusion.

This Has Happened Before

None of this is new. We have actually run a pretty close to the T full dress rehearsal of the 2026 card panic, complete with congressional actors and death-of-the-network headlines.

In 2010, the Durbin Amendment came out of the Dodd-Frank Act (rest in peace, Barney Frank) and did the thing today’s bears only theorize about: it actually capped fees by law, cutting debit interchange for big banks by nearly half when the Fed’s caps took effect in October 2011. The day-of coverage read exactly like this year’s: the government had finally come for the card industry, the economics were broken, the growth story was over. Visa fell 12.7% in a single session — December 16, 2010 — when the Fed’s proposed caps came in harsher than expected, and Mastercard dropped 10% alongside it.

What actually happened next: the issuers ate the cap, debit rewards died, a few banks tried $5 checking fees and got publicly tarred and feathered for it, and the networks — whose tolls were never the thing being capped — went on the best decade of their corporate lives. From the last close before the caps took effect through the end of 2021, Visa returned roughly tenfold (10.1x) on price alone, before counting their dividend.

The lesson isn’t that regulation doesn’t matter. It’s that regulation aimed at the card ecosystem keeps landing on the issuers and the rewards programs, while the rails underneath keep collecting tolls on a payment volume that grows through everything, because the alternative to the rails has to be built, trusted, distributed, and defended against fraud — and nobody who has tried has survived the attempt.

And people have tried. The obituary file is thick. E-commerce was going to cut out the networks; instead, card-not-present volume became their growth engine. Mobile wallets were the next assassin — if 18-year-old Sterling remembers the 2014 coverage of the Apple Pay announcement correctly, it was looked at as an extinction event and the dawn of a brave new world — but Apple Pay turned out to be a client of the payment railroad, a beautiful new front door that pays the same toll. Bitcoin and the rest of Crypto was going to do it in 2021; the networks now settle in stablecoins and charge for the privilege. The pattern is twenty years old: every payment innovation begins as the networks’ replacement and ends as their customer. The 2026 panic requires this pattern to break for the first time. Maybe it does! But that’s the actual bet the naysayers are making, whether they say it out loud or not.

Real Fears

I’m not going to strawman it, because pieces of it are real, and one piece genuinely matters.

Stablecoins. The number you keep seeing is $35 trillion of stablecoin transfer volume in 2025, up just over 70% year over year. That number is doing dishonest work in every bearish deck it appears in. Strip out the trading settlement, the exchange churn, the bot wash — which McKinsey and Artemis have both taken a scalpel to — and genuine payments on stablecoin rails were roughly $390 billion last year, about one percent of the headline number. Card-linked consumer stablecoin spending was around $4.5 billion. Visa and Mastercard’s combined annual purchase volume is measured in the tens of trillions. So at the point of sale, today, the disruption is a rounding error on a rounding error — and the networks are responding the way incumbents with 50% margins can afford to: buying the threat. Mastercard agreed to pay up to $1.8 billion for BVNK, a stablecoin infrastructure firm. Visa, Mastercard, and Stripe are reportedly standing up a shared stablecoin platform. Where stablecoins genuinely bite first — cross-border B2B, remittances, treasury flows — the networks intend to own the toll booth on the new road too. The honest version of the stablecoin risk isn’t this year or next year’s volume; it’s whether a merchant coalition with real distribution ever gives consumers a bribe big enough to leave the cards. Hold that thought for the bear section.

Washington. This regulation threat is across three separate fronts. The merchant interchange settlement — ten basis points off average effective credit interchange for five years, a 1.25% cap on standard consumer cards for eight — received preliminary approval from a Judge earlier this week, which converts a twenty-year litigation tail into a known, bounded number that lands mostly on issuer economics (the interchange merchants pay goes overwhelmingly to the card-issuing banks, not the networks — the networks’ own fees are a separate, much smaller line). The 10% rate-cap idea is now an actual bill, S.381, sitting in committee; read it and notice what it regulates — lending — and which of our three contestants is a lender? And the Department of Justice’s monopolization case is aimed at Visa’s debit business by name, with fact discovery running to October 2026 and experts into 2027: years of hostile headlines, all of them spelling V-I-S-A. The composite picture is the Durbin rhyme again — every government gun pointed at the ecosystem is aimed at somebody’s piece of it, and the piece that is hardest to even theorize a remedy against is the pure network toll.

AI agents. The February note from Citrini Research that lit this fuse argues that agents optimized for transaction cost will route around 2–3% card rails toward rails that cost basis points. As a provocation it’s great. As economics it repeats the interchange confusion — the 2–3% isn’t the networks’ fee — and it skips the part where an autonomous agent spending your money needs more trust infrastructure than you do: authentication, authorization scopes, disputes, fraud screens, recourse. Somebody will be paid to provide that, and both networks are already building it — Visa Intelligent Commerce and Mastercard Agent Pay both shipped as programs over a year ago, tokenized agentic-commerce frameworks where the agent carries a credential the merchant can verify. The agent future is a fee opportunity wearing a disruption costume, unless the trust standard gets built by someone else — which is a real tail, and it’s in the bear section too, where it belongs.

The Race: Visa, the Empire

Visa is the biggest, the oldest story, and right now the best performer of the three, down only 7.3% on the year while the other two are down thirteen-plus — the market has already noticed some of what this piece argues. It started as BankAmericard in 1958, a Fresno experiment that became the first widely successful general-purpose credit card and, through licensing, the first national bank-card network — renamed Visa in 1976 — and it has been the default rail of global commerce more or less ever since. Forty billion of fiscal-2025 revenue, half of it net income, the deepest cross-border franchise in the world, and a fiscal second quarter that was its best revenue growth in four years — plus a brand-new $20 billion buyback that says management’s read of intrinsic value matches mine.

The case for just buying Visa is real: most scale, fattest margin, and at 24.7x FY26 consensus, the multiple has rarely been this forgiving. If you want one-decision exposure to everything in this piece, nobody will fault you.

It finishes second anyway, for one structural reason and one legal one. Structurally, Visa is the most exposed of the three to US debit — the slowest-growing, most regulated, most commoditized corner of the industry; the DOJ’s own complaint puts Visa at more than 60% of US debit transactions against Mastercard’s roughly 25%. Legally, that same franchise makes Visa the named defendant in the one piece of litigation that targets a network rather than the ecosystem. I think the remedy risk is overpriced and the headlines are worse than the math. But in a race between near-identical toll roads, you don’t volunteer for the one with the docket. You’re getting Visa’s discount because you’re getting Visa’s docket.

The Race: Amex, the Survivor

American Express is the most interesting wrong answer I’ve looked at all year.

Understand what it actually is: not a toll road but a club that owns its own road. Amex issues the cards, extends the credit, runs the network, and signs the merchants — the closed loop — which means it keeps the discount revenue and the annual fees and the net interest income, and it eats the credit losses and wears the lender’s regulatory target on its back. It is also, culturally, the great survivor of American finance: the company is 176 years old, was a freight-forwarding business before the Civil War, and has been declared doomed roughly once a decade since. The 1963 salad-oil scandal nearly broke it, and a young Warren Buffett famously put up to 40% of his partnership’s capital into the wreckage at the position’s peak, because he noticed people were still using the card; Berkshire owns more than a fifth of the company to this day — 22.2% at last count. Every era’s disruptors — Diners Club, Discover, fintech, BNPL — took their shot, and the green card outlived them all. Betting against Amex has been one of the most reliably losing trades in the market for a century. I’m not making it now.

What I’m making is a ranking. And the 2026 panic, decomposed honestly, lands its only real punches on Amex’s model. The rate-cap bill regulates lending: Amex lends. The credit cycle is the one disruption that requires no innovation at all: Amex books the provisions. The premium-consumer concentration that makes the franchise gorgeous in expansions has never been tested with today’s Gen-Z-and-millennial-heavy cardmember base in a real downturn — and the company is winning those younger cohorts precisely by front-loading rewards spend against lifetime values that a recession would re-mark. Q1 was excellent — EPS up 18%, billed business up 10%, guide reaffirmed at $17.30 to $17.90 — and at 18x the midpoint of its own guide, it’s the cheapest of the three. It’s also the only one of the three whose cheapness has a conventional explanation: lenders trade like lenders. The market isn’t mispricing Amex. It’s pricing it correctly and mispricing the other two. Third place, with affection.

The Race: Mastercard, the Operator

Now the winner.

Start with growth, because it’s the cleanest fact in the piece: Mastercard is the fastest. Fiscal 2025: net revenue $32.8 billion, up 16%, adjusted EPS $17.01, up 17% (15% currency-neutral). First quarter of 2026: revenue up 16% again, adjusted EPS up 23%. This isn’t a one-year sprint — Mastercard has out-grown Visa in seven of the past ten years, and the three years Visa “won” each came with an asterisk: an acquisition year and two fiscal-calendar quirks around COVID. Partly that’s structural geography (more Europe, where cash conversion still has room; less US debit, where growth goes to die), partly the thing I’m about to spend two paragraphs on.

The thing is the second engine. Mastercard’s value-added services — fraud and security tooling, authentication, data analytics, consumer engagement, the unsexy software layer of moving money — grew 22% in the first quarter and now make up 41% of total revenue. Sit with that number, because the entire bear thesis runs through rails, and two-fifths of this company is not a rail. It’s software sold on top of rails — anyone’s rails. New payment forms don’t shrink this business; they feed it. A stablecoin transaction needs fraud screening more than a card swipe does, not less. An AI agent needs identity verification more than a human does, not less. Mastercard built the one large business in the payments industry that is structurally long payment chaos, and it grew faster than the network itself for years while the market was busy watching the rails. When I said the panic can’t touch the best part of this company, this is the part I meant.

Then run the regulatory ledger from earlier, contestant by contestant. Lending exposure: Amex yes, Mastercard no. DOJ defendant: Visa yes, Mastercard no. Outsized US debit franchise under Durbin’s regime and the DOJ’s eye: Visa yes, Mastercard a quarter of it. Interchange compression: lands on issuers, not networks. Every fear on the 2026 list, traced to an actual P&L line, lands hardest somewhere other than Purchase, New York. Mastercard isn’t immune — nothing with a 40-handle margin is ever truly off Washington’s menu, and I’ll get to contagion below — but in a panic that prices all three animals as the same animal, it owns the shortest list of ways to get hurt, the fastest growth, and the biggest non-rail business. That’s the trifecta.

And here’s the kicker the market handed us somewhere during the selloff: the famous Mastercard premium has almost vanished. For most of the past decade you paid up meaningfully for the faster horse. Today Mastercard trades at 25.2x FY2026 consensus earnings against Visa’s 24.7x — half a turn. Against its own history the de-rating is more dramatic: this stock has averaged roughly 36x trailing earnings over five years and hasn’t finished a year below about 34x; it sits at 29.1x trailing today, on 2025 adjusted EPS of $17.01. The market is offering the best house on the street at the price of the street. I’ve spent four pieces explaining why some cheap stock deserved a premium. This is the first time I get to write the lazier, better sentence: the premium asset stopped costing a premium, and nobody adjusted.

The Math

From $17.01 of 2025 adjusted EPS, the Street compounds Mastercard to about $27 by 2028 — $27.26 per WallStreetZen’s consensus compilation as of June 2026, and worth knowing: analyst coverage that far out is thin, so treat the 2028 number as a single compiler’s estimate. The path implied is mid-teens earnings growth, which is not a hockey stick; it’s this company’s resting heart rate. Then it’s a multiple argument, and I’ll argue it from loser to winner here:

• Bear: 22x. A multiple this stock hasn’t seen since 2012, awarded only if the disruption narrative wins the argument without ever showing up in the numbers. 22 × ~$23 of haircut earnings ≈ $505 — two percent above today’s price (whatever the price was when we started the piece). Read that again: the bear case, on a bear multiple and bear earnings, roughly returns your money. The growth pays you to be wrong.
• Base: 29x. Still below anywhere this stock has finished a year in half a decade. 29 × ~$27 ≈ $780, about 57% upside, high-teens-to-twenty annualized into late 2028.
• Bull: 34x. The bottom of its own historical range back, on the same earnings: ≈ $920, if the panic simply ends and the multiple merely stops being insulted.

That asymmetry, which is flat on the downside scenario and +57% on the base case one, is the trade. It exists because the market is charging you nothing for the growth gap and nothing for the second engine. I’m not paying a premium and arguing it’s earned, like I had to with Meta’s options ladder or Nvidia’s margins. I’m being handed the premium asset at the field’s price during a panic about somebody else’s fees.

What Could Go Wrong

Six ways, ranked by how much they actually worry me — and the first two worry me for the same reason: they require nothing unusual to happen.

One: the growth gap closes. The whole structure rests on Mastercard out-growing Visa by several points a year. That gap is a fact with a decade of history, not a law of physics. Cross-border normalizes. Europe matures. The Middle East drag the company itself flagged in its second-quarter guide spreads. VAS comps harden off a 41% base. If Mastercard prints two quarters of Visa-grade growth, I own the formerly-premium horse at parity pricing and the market re-asks why it ever paid up — that’s a 20% air pocket with no villain and no headline. This is the bear case I check every quarter, because it’s the only one that arrives silently.

Two: the multiple regime changed and the history is a trap. Every scenario above leans on “hasn’t finished a year below 34x” — and hasn’t is not can’t. I wrote a whole piece about Lululemon, where a decade of 30-something multiples turned out to be a regime, not a floor, and the floor was 9. If the market has permanently re-filed payment networks from “software-margin compounders” to “terminal-value-question-mark financials,” then my anchor is nostalgia and the base case is the bull case. The tell to watch: what the stock does on good prints. If Mastercard beats and the multiple doesn’t respond — the way it hasn’t, candidly, for two quarters — the regime question is live and this thesis is on a clock.

Three: the checkout button arrives. Today’s stablecoin retail volume is a rounding error and the networks are buying the infrastructure. The break-glass scenario isn’t crypto-native — it’s distribution-native: a Walmart-and-Amazon-grade coalition shipping a consumer wallet with a bribe big enough to matter, instant and automatic, funded by the interchange they’d skip. Consumers have refused every previous invitation to abandon cards because the bribe was small, the protections were worse, and the habit was strong. Behavioral moats are the strongest kind right up until the day they aren’t. Low probability per year; terminal-value severity; un-hedgeable except by watching merchant-coalition news like it’s a geopolitical feed.

The next three are less likely, but they carry more shock if they land. I rank them as follows.

Four: Washington stops aiming at other people. My argument is “shortest list,” not “no list” — and lists grow. A DOJ win on Visa’s debit routing becomes a template applied network-wide; the Credit Card Competition Act, the one proposed law aimed at network routing itself, comes back with momentum; the settlement dies on appeal and gets replaced by something legislated. Durbin is one data point, and “the last bullet missed” is not armor.

Five: the robots really do defect. I hold my trust-infrastructure dismissal with one hand. The other remembers that B2B payments quietly standardized on cheap, slow, cardless ACH once software made the rails invisible. If the agentic trust standard is built by someone other than the networks, the tolls go elsewhere, and the volume I’m underwriting migrates at machine speed, not consumer speed. Too early to price; too coherent to ignore.

Six: I’m right about everything and Amex still wins. This is a horse race; I’m graded on the order of finish, not just the direction. If no recession arrives before 2028, the cheapest horse with the lending leverage — 18x guide, EPS compounding 18% — out-runs the safest one. I’m not paid enough to underwrite “no recession before 2028.” Still: the scenario where my ranking is wrong is more likely than any scenario where the thesis is wrong.

Conclusion

We ran this exact experiment in 2010, with an actual law instead of a hypothetical one, and the rails returned tenfold while everyone was reading the obituaries. Visa gives you the empire at a fair discount, with the docket stapled to it. Amex gives you the deepest discount on the one model the politicians are genuinely aiming at; correctly priced, wrong instrument. Mastercard gives you the fastest growth in the group, a second engine that is forty-one percent of revenue and feeds on payment chaos, the shortest regulatory list — and, for the first time in roughly a decade, all of it at the field’s multiple.

For the first time that I’ve written one of these, I don’t have to end by saying I can’t believe a giant company is cheap. Mastercard isn’t cheap. It’s mispriced relative to itself — 29x trailing against a history that never let it finish a year under 34, with the bear case math working out to roughly my money back. The panic is pricing the death of the toll roads. The toll roads are reporting the best numbers they’ve printed in years and I’ll take the fastest one.

view all sources here

Suggested Citation: Rettke, Sterling. “Priceless: $MA Is Best in Class.” sterlingrettke.com, June 12, 2026.

Disclosure: I do not currently hold a position in $MA, $V, or $AXP, though I look at all three from time to time. This piece is for informational and educational purposes only and is not investment advice. Do your own research.