Affirm is a payment network, not a lender, investor argues

1. A bad credit score — ▶ 2:27

   Levchin left Ukraine after Chernobyl, then used credit cards to fund university startups and paid late, damaging his credit. After the PayPal IPO, a Los Angeles car dealer declined his loan, and he waited for a wire to clear. The mission that followed: no late fees, no compounding balances, no deferred interest. Industry insiders thought this insane because the sector's profits depend on exactly those mechanisms. Christie predicts that in 5-10 years Levchin will be known as Affirm's founder first, with Affirm proving more influential than his PayPal chapter. That is a forecast, not an outcome.

2. Hard problems, flat costs — ▶ 5:34

   Levchin screens ideas for being hard, valuable and fun. Affirm chose the hard version of lending: large unsecured amounts over long periods, without late fees as a safety valve. Christie sees the same instinct in cost discipline. Once among the heavier stock-compensation offenders after listing, Affirm has held headcount flat for years, G&A flat for about 5 years and stock-based compensation flat for about 6. Christie says it now reports compensation levels in line with the Magnificent 7, which he calls middle of the road.

3. Anatomy of a transaction — ▶ 7:42

   On a typical $250 purchase, theoretical offers appear first, for example 0% over 3 months or 25% over 12 months, and are finalized only after Affirm matches basket and merchant data with the buyer's credit data. The rate and schedule are fixed and the balance runs to zero. Affirm pays the merchant the price less a merchant discount rate (a $5 example), and earns interest when the loan bears it. About half of loans are sold off balance sheet. Christie contrasts this with cards: half of US cardholders revolve, with an average balance of $7,000.

4. Underwriting every transaction — ▶ 12:10

   A card issuer grants a general-purpose line and leaves the customer to manage it. Affirm re-prices and re-sizes every purchase against changing rates, economic conditions and basket contents. Management says new transformer-based models arrive every 6-7 months, each materially better. Christie's comparison with Synchrony: roughly 75% of Synchrony's customers are prime against about 50% of Affirm's, yet Affirm reports better underwriting outcomes. These are company claims and Christie's benchmarking, not audited comparisons, but he treats transaction-level underwriting as a structural step change that models then compound.

5. Not a subprime lender — ▶ 15:14

   Christie calls it a misconception that fintech lenders mainly serve non-prime borrowers. By dollars lent, Affirm's average customer is generally prime, and the base skews 30 and older. The target is the large share of cardholders carrying revolving balances, not Amex Platinum holders chasing lounge access. Because amounts and terms are tuned per person, Affirm can approve across more credit-score bands, such as a $100 short-term loan a card would not extend, while reporting strong repayment and low delinquency. This also widens merchants' addressable shoppers.

6. Interest-bearing loans earn more — ▶ 17:26

   Roughly 70% of GMV is interest-bearing, with the rest split evenly between 0% loans and pay-in-X. Pay-in-X earns revenue less transaction costs of about 1-3% of GMV, mostly from merchant fees now approaching card rates. Interest-bearing loans earn 4-6%, mostly interest, after losses and funding costs. Blended, Affirm earns about 4%, which Christie calls industry leading. The feared dilution from the Affirm Card has not appeared: card volume skews even more heavily toward interest-bearing loans, and economics there are similar to direct merchant integrations.

7. Network effects in the data — ▶ 21:28

   Unlike Visa and Mastercard, which only provide rails, Amex integrates merchants and credit; Affirm does both online. Christie's evidence: growth above 30% at over $50 billion, faster than a few years ago; usage up from 2 transactions per user per year at IPO to 7; about 95% of quarterly transactions from existing customers; and cash sales and marketing spend barely changed since 2021. Apps that load credit onto a virtual card lack this loop and are lenders. The claim of negative CAC excludes equity issued to Shopify and Amazon.

8. BNPL at 6% of e-commerce — ▶ 26:11

   E-commerce is about 20% of US retail sales, and BNPL is about 6% of US e-commerce versus 15-25% in Australia, Sweden and other European markets, where Afterpay and Klarna originated. Christie argues US BNPL penetration can triple. Affirm holds about 50% US share, is gaining, and is growing faster at its scale than Klarna or Afterpay. Network effects, he argues, make a leader with escape velocity hard to catch, because merchants choose the button with the largest customer base. The share and penetration figures are his estimates.

9. Agents favor subsidized credit — ▶ 29:03

   Financial stocks fell on the idea that agents will compare offers better than people. Christie accepts that but argues Affirm benefits: it sits at checkout with personalized offers, no late fees and merchant-subsidized 0% loans on items like a $250 couch, which no card can match. An agent can pre-fill the application, reducing friction. He expects Affirm to supply white-label credit to agent wallets and share economics, as with Shop Pay installments. This is his expectation, not demonstrated revenue; agentic commerce is excluded from management's growth target.

10. Hard loans first — ▶ 32:39

    Affirm is more than 2x Klarna's size in the US, and Christie's explanation is product sequencing. Affirm began with large, long-dated loans, such as Peloton financing over 24-36 months, then backfilled easier ones. Its monthly installment cohorts show loss rates around 3.5%, versus about 1% on short, low-ticket loans. A $500 loan over 12 months gives a consumer 12 chances to decide not to pay. Klarna, he says, started with easier loans, is moving longer, and still earns from late-fee mechanisms Affirm has renounced.

11. Card, merchants, Walmart — ▶ 38:32

    Only 75 of the top 250 US merchants have Affirm integrated. Adoption is slowed by merchants treating payments as a cost; once Amazon integrated, others followed quickly. The Affirm Card, on the Visa network, is about 20% of GMV, growing over 100%, and its data on spending at non-integrated merchants becomes a sales tool. Walmart switched to a Klarna-powered offer, but Affirm says impact was minimal: Walmart was low-profit, and a large majority of its customers kept shopping there using the Affirm card. What Affirm loses is the acquisition channel.

12. From $9 to $60-80 — ▶ 46:30

    Affirm listed at about $49, closed day one near $100, hit about $175 a year later, then fell to $9, a drawdown above 90%. Christie attributes most of it to the multiple compressing from 20 times revenue to 20 times earnings. The business itself deteriorated for one quarter, December 2022, when growth slowed and margins fell, before recovering in the next two. Now it trades at $60-80, about a $25 billion fully diluted market cap and about 20 times adjusted earnings excluding stock compensation, on roughly $50 billion of GMV and about $2 billion of revenue less transaction costs.

13. Alignment and AI culture — ▶ 51:53

    Levchin's salary is under $40,000, with no cash bonus, while Christie says he owns a couple of billion dollars of stock. The two presidents, each 8-10 years at the company, own more than $100 million of stock each. Christie says 70-80% of code is already deployed by AI. Levchin now spends more time in an innovation lab, from which the Affirm Card, the merchant-funded 0% network and Edge (white-label Affirm for regional banks) emerged, while the presidents run the core business. Christie calls the comparison with other card businesses unfair, an opinion without comparative data.

14. Rates cause air pockets — ▶ 56:56

    Funding is diversified: about 50% of loans off balance sheet, 50% in warehouses and revolving ABS, with a banking license in process. Loans carry fixed rates over terms of about 11-12 months, so after a hike roughly 50% of the book reprices within a quarter and 75% within two. Management cites about $90 million of P&L impact from a 100 basis point rise against about $2.5 billion of gross profit this financial year; a 500 basis point shock is the real problem. The main effect is slower growth as Affirm stops writing marginal loans, not a surge in defaults.

15. $100 billion excludes upside — ▶ 1:03:43

    Against $1.3 trillion of US card balances and about $13 trillion of annual Visa, Mastercard and Amex volume, Affirm's roughly $10 billion loan book and $50 billion of GMV are small; Capital One processes over 15x Affirm's GMV. Management targets $100 billion at above 25% annual growth, from core US, card and international, with Edge and agentic commerce excluded. International matters strategically: Airbnb and eBay chose Klarna for a single global button, and Christie says Affirm is being pulled abroad by Shopify and Costco.

Fraser Christie's central claim is that Affirm belongs in the same category as American Express rather than Capital One or Synchrony. His distinction is structural. Visa and Mastercard are rails only; card issuers sit on top and extend the credit. American Express integrates both, holding direct merchant relationships and the lending. Christie argues Affirm has done the same thing online: merchants install its checkout button, and Affirm underwrites the loan behind it. Amex has always traded at a discount to Visa and Mastercard because of its credit component, and Christie accepts that Affirm is valued like a lender today. His argument is that the network deserves a premium multiple, and that investors keep debating whether Affirm is a lender or a tech business when the answer is both.

The test of a network is whether growth compounds with scale, and here Christie offers his most concrete evidence. Affirm processes more than $50 billion of GMV, growing over 30% a year, faster than a few years ago despite the business now being two to three times the size. Usage per customer has risen from 2 transactions a year at IPO to 7, and about 95% of transactions in any quarter come from existing customers. Cash sales and marketing spend has barely moved in dollar terms since 2021. The mechanism is circular: more merchants expose more shoppers to offers, and more shoppers make the button worth installing. Apps that load credit onto a virtual card lack that loop, in his view, so he calls them lenders.

Several qualifications temper this. The negative customer acquisition cost Christie describes holds for the average merchant, but not for Shopify and Amazon, to which Affirm issued equity when it signed the original deals. Concentration is real: the host noted that Amazon alone is over 20% of the business, and Christie concedes that losing Amazon or Shopify would hurt the acquisition flywheel. His comfort rests on long-term agreements and his understanding that Levchin and Shopify's leadership are close. The Walmart departure is the live test. Walmart switched to a Klarna-powered offer, yet the impact was minimal because Walmart was a low-profit merchant and, per the company, a large majority of its customers kept using Affirm through the card.

A network needs a credit engine that can survive the volume. Affirm's differentiator, Christie says, is underwriting each transaction individually, matching merchant, basket and SKU details with credit data, rather than granting a general-purpose line and leaving customers to manage it. Management says new transformer-based models arrive every 6-7 months, each materially better. Christie's benchmark is Synchrony: roughly 75% of its customers are prime against about 50% of Affirm's, yet Affirm reports better outcomes. These are company claims and one investor's comparisons, not audited findings, but the structural point is coherent. Because each loan is sized and priced for a specific purchase and designed to amortize to zero, the product is a bounded commitment rather than a revolving balance.

That bounded structure explains where Affirm earns its money. About 70% of GMV is interest-bearing, with the rest split evenly between 0% loans and pay-in-X. Pay-in-X earns 1-3% of GMV in revenue less transaction costs, mostly from merchant fees. Interest-bearing loans earn 4-6%, mostly from interest, after losses and funding costs. Blended, that is about 4%, which Christie calls industry leading. The implication is that the harder product is the better one, and it is what Affirm built first. Its monthly installment cohorts show loss rates of about 3.5%, against about 1% on short, small-ticket loans, and Klarna, in Christie's telling, began with the easier loans and is now moving toward longer ones.

Christie spends time dismantling the idea that this is subprime lending. By dollars, Affirm's average customer is generally prime, and the base skews 30 and older. The target is the half of US cardholders carrying a revolving balance, with an average card balance of $7,000, not the Amex Platinum holder chasing lounges. Levchin built the company on no late fees, no compounding and no deferred interest, mechanisms that industry insiders thought a lender could not live without. The merchant side adds a further wrinkle. A merchant-funded 0% loan can cost the merchant a high single-digit fee, but Affirm pitches it as personalized discounting versus a 20% storewide sale, a pitch some merchants still resist.

The growth case rests on three layers. E-commerce is about 20% of US retail sales, and BNPL is about 6% of US e-commerce versus 15-25% in Australia, Sweden and other European countries; Christie thinks US penetration can triple. Affirm holds about 50% US share and is gaining. And only 75 of the top 250 US merchants have Affirm integrated, while the Affirm Card, now about 20% of GMV and growing over 100%, produces data on where customers shop at non-integrated merchants. Management targets $100 billion of GMV at above 25% annual growth, from about $50 billion in the year to June, covering core, card and international. Edge, a white-label offering for regional banks, and agentic commerce sit outside that target. Against $1.3 trillion of US card balances, the runway is large.

Agentic commerce is where the bull and bear cases collide. Financial stocks sold off on the idea that AI agents will compare offers better than humans. Christie accepts the premise and flips the conclusion: an agent weighing a revolving card balance against a merchant-funded 0% offer on a $250 couch should pick the latter, and it can complete the application without friction. Affirm already supplies white-label credit for Shop Pay installments, and he expects the same arrangement with agent wallets. Two caveats are editorial inferences. The argument depends on merchants continuing to fund 0% offers, which Christie says some find excessive. And agentic revenue is so far a hope, since it is excluded from guidance.

Interest rates are the mechanism the market fears most, and Christie's explanation is more specific than the usual worry. Affirm funds loans through a deliberately diversified base: about 50% off balance sheet in forward flow and static ABS facilities, and 50% on balance sheet in warehouses and revolving ABS, with a banking license in process. Loans carry fixed rates over terms averaging 11 or 12 months, so after a hike roughly 50% of the book reprices within a quarter and 75% within two. The result is a one-to-two-quarter air pocket. A 100 basis point rise would cost about $90 million against roughly $2.5 billion of gross profit this financial year. The larger effect is slower growth, as Affirm stops writing loans that no longer clear its risk threshold, rather than a wave of defaults.

The 2021-2023 share price shows how that risk reads in the market. Affirm listed at about $49, closed the first day near $100, reached about $175 a year later and bottomed at $9, a drawdown of more than 90%. Christie attributes most of the fall to the multiple moving from 20 times revenue to 20 times earnings. The business itself stumbled for one quarter, December 2022, when growth and margins deteriorated sharply, then recovered over the next two. Today the shares trade at $60-80, a fully diluted market cap of about $25 billion and about 20 times adjusted earnings excluding stock compensation, on about $50 billion of GMV and about $2 billion of revenue less transaction costs. Christie reads the recovery as evidence the business self-heals.

That history also explains the valuation tension. Christie says growth investors, uncomfortable with credit, exit whenever a headline suggests private credit is impairing, while bank-style investors arrive in weak periods but do not value a founder-led tech culture. The stock therefore oscillates between a 10 times and a 20 times earnings argument. The most common objection he hears is not Affirm-specific but discomfort with credit businesses in general, plus the subprime misconception, which he compares with how Australian investors dismissed Afterpay. His own circle may be biased, he suggests, because they optimize for points and lounge access, not revolving balances.

The alignment argument is the last pillar. Levchin's salary is under $40,000, he takes no cash bonus, and Christie says he owns a couple of billion dollars of stock. The two presidents, each with the company 8-10 years, own more than $100 million of stock each. G&A has been flat in dollars for 5 years and stock compensation for 6. Christie says 70-80% of code is already deployed by AI, and Levchin now spends more time in an innovation lab, where the card, the merchant-funded 0% network and Edge originated, while the presidents run the core. Christie's assertion that other card businesses cannot compete with this culture is his opinion; he offers no comparative data.

The unresolved issue is what would change the network-versus-lender verdict. Growth, acquisition costs and underwriting results all support the network reading, but the evidence comes largely from company disclosures and one investor's comparisons. The cleanest test would be a period in which funding markets tighten and consumers weaken together, while agent-mediated checkout pressures merchant-funded 0% subsidies. The 2022 episode showed Affirm can absorb an air pocket. It did not show how the model behaves when losses, rather than growth, become the main variable.