There Are No Sub-$100B Niches in Payments
Payments is the biggest market in the world — any slice you think is narrowest still pencils out to $100 billion. But the bigger the amount, the smaller the take rate; the real money is in small-ticket, high-frequency.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
There are no sub-$100B niches in payments
Max Levchin says payments is ‘by any imagining the largest market in the world’ — any slice you think is narrowest and most contrarian still pencils out to $100 billion. But Alex Rampell adds a counterintuitive mechanism: once the amount gets very large, the take rate gets very small. You can't charge 2% on a $40 trillion transfer. So large-ticket payments have enormous volume but are often not large revenue opportunities; the real revenue opportunity is at the small-ticket, high-frequency end. The one exception is B2B payments — everyone thinks they have a clever B2B payments idea, and it happens to be the exception to this rule.
— Max Levchin / Alex RampellThe Visa network's hard limit is 2.5 seconds
Max points out that Visa and MasterCard have a hard 2.5-second limit on the interaction between the network, the issuing bank, the merchant and the acquirer; past that, the transaction gets retried or simply cancelled. That means card-present payments leave almost no room for any ‘clever innovation’ — online commerce can run fraud checks before submitting the card number, but once the card is handed over in person you have only 2.5 seconds. Apple Pay and Google Pay matter because they built a secure enclave into the chip, doing the ‘I already know your card’ part in advance and thereby moving the whole timeline out of the way. What genuinely surprised Max: Visa and MasterCard still haven't put out a new standard saying ‘actually it doesn't have to be 2.5 seconds, it could be 15 seconds, let's go get a bunch of issuing banks to bid you better credit terms.’
— Max LevchinPayment innovation has no ‘pretty good’ ending
Max tells a failure he saw with his own eyes: before PayPal, MasterCard and a gas station network built a ‘magic wand’ you could wave next to the pump to fuel up and pay. He thought at the time, ‘God, this obviously replaces the credit card.’ It didn't. The rule he drew from it: payment innovation has a critical mass that isn't obvious ex ante but is obvious ex post — miss it and you die, there is no ‘pretty good’ middle outcome. Either everyone has to have your widget, your network, or you go on the scrap heap of history. The wand was only a little faster than the credit card in your pocket, not a lot faster, and the credit card just works. Amazon also killed palm payments at Whole Foods; Max says he loved it and used it every time he stayed at a hotel, but it was actually slower — it was just fun.
— Max LevchinThe pajama problem and ‘paying with your identity’
Affirm's starting point was a concrete image: you're upstairs in your pajamas, you want to buy something, and your credit card is downstairs — how do you pay? Alex's reasoning: phones are spreading, social networks are spreading, so maybe there could be a new credit-granting mechanism, a return to the 1800s general store — you don't have cash on you, but the shopkeeper knows you and says ‘Max, I trust you, settle up next time.’ Today, in a Walmart store, the greeter has no idea who you are and can't do the general-store thing; but if you have 500 friends on Facebook and have uploaded a thousand photos, you're probably a low credit risk. The key is that ‘you're not looking for credit’ — people actively seeking credit (Googling ‘broke, need money, credit’) are often bad credit risks, whereas the general-store model serves people who aren't looking for credit, they just didn't bring their wallet.
— Alex Rampell / Max LevchinIt took a 30% conversion lift to find PMF
Affirm's stretch with 1-800-Flowers never went well: Amit was never happy with the conversion rate, and said flat out, ‘You're cannibalizing my credit card volume and charging more than the credit card — this is a dumb idea.’ The turn came from Beautylish — the only difference was that it moved the line ‘you can pay in three installments’ upstream in the shopping flow, telling the customer while they were picking out shampoo and perfume, and conversion immediately rose 30%. Max says that was the moment they understood: this isn't an alternative payment method, and it isn't solving the pajama problem — it's solving ‘my budget is only this much, but if I can split it up, my budget gets bigger.’ They then turned that playbook into a sales offensive, finding a batch of small DTC brands that didn't care whether they were charged 1%, 5% or 12%, because what those people wanted was to grow revenue.
— Max LevchinEvery mattress company read the same HBR article
Max says the Harvard Business Review article from around Casper's founding was read by all of these founders: people buy a mattress once every seven years, and if you're the company selling that mattress, this is the most important thing in the world — miss it and you wait another seven years — and mattress gross margins are astonishing. So within four or five years a pile of companies appeared, with the logic: either get you to shorten the replacement cycle from seven years to six, or get you to decide this mattress is the best, coolest one and it springs out of a box. The result was a saturated mattress industry, but it also created a huge premium on ‘get them to buy a mattress at any cost’ — ‘I'll give you a three-year zero-interest loan, provided by Affirm’ became a very worthwhile marketing cost. Alex adds that mattress margins are so high that tolerance for MDR is much greater, and nobody wants to pull out money for a $1,200 mattress while in their pajamas — once you split it into 30 installments, conversion explodes.
— Max Levchin / Alex RampellReal zero interest is anger at fake zero interest
Max says he will stay angry until this product is legislated out of existence: the ‘open a store credit card, zero interest’ sign at a department store, where the zero has an asterisk meaning it assumes you keep spending over the next 12 or 24 months; if you pay a penny less or a day late, interest is retroactive to the day you got the card — you spent $1,000 and wake up two years later owing $3,000. It's called a deferred interest credit card. Affirm's real zero interest comes directly from that anger: a month late, a year late, the price doesn't change, and Affirm's zero never carries an asterisk. That's also the origin of their not charging late fees and not doing deferred interest — ‘you decide to do three years with us, you will never get screwed, you will never get a negative surprise.’
— Max LevchinAffirm's customer acquisition cost is negative
Alex says that as a VC, looking at 90% of consumer companies, he'd rather buy Google or Facebook stock, because those companies buy their customers from them. And the coolest thing about Affirm in his eyes: it has a negative CAC — it gets paid to acquire a customer. The mechanism is B2B2C: Casper, the friendly white-ghost mattress company, doesn't want to personally send you a dunning notice saying ‘get off my mattress, you haven't paid me back’; it wants a third party to do that. Alex says he tried B2B2C at TrialPay, but Zynga and Netflix didn't want him owning their customers; Affirm's merchants, by contrast, want Affirm to own the customer, because supporting a product, handling unhappy customers and tech support is already a burden — a one-year loan needs 12 reminders, a three-and-a-half-year loan needs 39 — and merchants are happy to hand off that communication burden.
— Alex RampellAgents will buy goods, but won't pick your clothes
Max says he's not that optimistic about agentic shopping, but very optimistic about agentic payments. The idea that ‘a robot will buy our Friday-night clothes for us’ is wrong — we want to know how something looks on us before the robot drops it at the door. But the credit card, ‘the best user interface ever built,’ may finally get renegotiated, because an agent really is smarter than a piece of plastic, or even a piece of plastic with a rewritable chip. Alex partly agrees but draws a boundary: when researching what to buy, AI is just another tool — you ask friends and you ask AI; but when you already know which SKU you want and only want the lowest price, that's where the agent truly takes over. He cites CamelCamelCamel — a top-100 US website, but people earning over $100,000 a year have probably never heard of it; it serves the ‘more time than money’ crowd.
— Max Levchin / Alex RampellIn their own words · checked verbatim
It is the world's largest market by any stretch of imagination, and there are no niches in payments that are smaller than $100 billion.
Max Levchin0:00
There are no niches and payments that are smaller than $100 billion.
Max Levchin6:08
There's not really a okay outcomes in payments. You either get there and everyone needs to have your widget or your network or your whatever. Or it's just going to go into the annul of time.
Max Levchin11:26
Basically, convenience just trumps everything else as the total amount you're trying to send goes down.
Max Levchin16:36
There will never be an asterisk on a firm zero.
Max Levchin40:37
I'm probably less optimistic about agentic shopping and I'm very optimistic about agentic payments.
Max Levchin52:52
maybe AI is already there. It's just that you haven't yet trusted your agent to do as good a job as you would.
Max Levchin56:56
Figures
| Visa/MasterCard network interaction hard time limit | 2.5 seconds | 4:04 |
| Minimum size of a payments niche | $100 billion | 0:00 |
| Conversion lift at Beautylish after moving installment messaging upstream | 30% | 33:08 |
| Mattress replacement cycle | once every 7 years | 39:34 |
| Number of Americans who have transacted with Affirm | over 50 million | 43:43 |
| Number of countries Affirm has entered | 4 | 43:43 |
| Number of payment reminders on Affirm's three-and-a-half-year loans | 39 | 46:47 |
| Number of payment reminders on Affirm's one-year loans | 12 | 46:47 |
| MDR Affirm quoted 1-800-Flowers early on | 7% | 30:06 |
| Date of the original email mentioned in the discussion between Affirm and Max | April 2011 | 7:09 |
Glossary
- MDR / merchant discount rate
- The rate a merchant pays to accept a payment, because cash has to be advanced to the merchant.
- APR / annual percentage rate
- The annualized cost a consumer pays to borrow, a separate thing from the merchant-side MDR.
- deferred interest
- A credit card term under which paying a penny less or a day late during a zero-interest promo period makes interest retroactive to the day of purchase.
- B2B2C
- A platform serves merchants first, and merchants then reach consumers; Affirm uses this to own the end customer.
- secure enclave
- An isolated secure area inside a chip; Apple/Google Pay use it to complete card verification in advance.
- factoring
- Selling receivables outright for cash; because it's a sale rather than a loan, it isn't bound by credit laws.
How to listen
Founders and investors in payments, consumer credit and DTC e-commerce, plus product leads who want to understand the BNPL economic model and the boundaries of agentic payments.
From 48:51, the reminiscing about why PayPal's early team produced so many founders — anecdotal, skippable.