The Legendary Debate

Warren Buffett: leans back, eyes twinkling, a gentle smile creasing his face “Banking has always been a peculiar animal. You're essentially selling trust wrapped in convenience and technology. So when we look at Nu Holdings, the question I keep asking is—does this outfit have an enduring advantage, or are we just looking at a fast horse in a crowded race? Everyone can lend money, everyone can build an app—but can everyone earn the kind of customer devotion NU seems to have among younger Brazilians and Mexicans? Is that loyalty durable, or just infatuation with low fees and slick UX?”

Phase 1: Business Quality & Long-term Moat

Charlie Munger: “To answer your question, Warren, I think what NU’s done is remarkable from a human behavior and cost perspective. Their moat is psychological as much as technological. They took a population suffocated by legacy bank bureaucracy and gave them simplicity, dignity, and speed—all at a fraction of the cost. That type of emotional moat can last a long time if continually reinforced. But here’s the inversion: easy onboarding breeds fickle customers. If someone builds a shinier app with 10% better rates, will they stay? Unless NU becomes the default financial operating system for consumers, we could be underestimating churn.”

Dev Kantesaria: “Charlie, that’s precisely my concern. NU clearly solved pain points in traditional banking, but the inevitability test fails. I look for businesses where success over a decade is unavoidable due to structural positioning. NU operates in a competitive, fast-evolving fintech universe. They have goodwill now, but goodwill is fragile when switching costs are microscopic and competitors like Mercado Pago and Itaú Digital Bank are fighting in the same trench. Even their AI lending models can be replicated by other players using public data. This doesn’t look like Visa or FICO—it’s too exposed to competitive mimicry.”

David Tepper: “Fair, Dev, but let’s not dismiss their scale advantage outright. 131 million customers and 83% active rate—that’s astonishing. In Brazil, they’re practically embedded as a national habit. I think the moat’s shifting from initial growth to data advantage. Millions of transactions feed their models daily, deepening predictive capabilities on credit risk. AI decisioning through nuFormer isn’t just hype—it’s real operational leverage. As Brazil’s macro stabilizes—lower inflation, easing rates—that data becomes monetizable through credit products at lower default rates. The network effects between engagement, lending, and deposits might be stronger than you're giving credit for.”

Robert Vinall: “I tend to agree with David on one point—the internal reinvestment dynamics here are unusually strong. NU isn’t paying dividends; it’s reinvesting nearly all free cash into new geographies like Mexico and Colombia, using Brazil’s compounding machine as an engine. This is classic Vinall-type compounding. But longevity depends on discipline—if each expansion market keeps its unit economics intact. The real question is whether NU can recycle cash flow at similarly high returns abroad. Many great franchises fail the ‘foreign transplant test.’ Coca-Cola succeeded because its product was universal. Financial trust is inherently local.”

Mohnish Pabrai: “Robert, I respect that argument, but the asymmetry isn’t good enough for me. NU’s success story is priced in. You’re betting on perfection across three different regulatory regimes. I’m not seeing margin for error. The fintech graveyard is full of businesses that looked like compounders until local politics or credit cycles broke the model. If a sharp recession hits Brazil and non-performing loans spike, all of today’s customer love can vanish fast. I need odds stacked 3-to-1 on the upside relative to downside, and this one feels maybe 1.5-to-1 at best.”

Pulak Prasad: “I’d echo that caution, Mohnish. Evolution doesn’t spare fintech. Remember M-Pesa in Kenya—revolutionary for a time, but multiple layers of competition and regulatory friction eventually dulled its edge. NU’s ecosystem may look entrenched now, but fintech is Darwinian. The environment keeps mutating—AI, payment rails, cross-border capabilities. A 20-year hold requires resilience beyond product innovation; it requires an institutional capacity to pivot before the terrain forces you to. I’m not sure NU has proven that yet.”

Charlie Munger: “Pulak’s making an important evolutionary point—survival bias fools people. If fintech economics deteriorate, empathy-based branding won’t save them. I like management’s rationality; David Vélez is clearly intelligent and culture-driven. But we shouldn’t mistake competence for inevitability. Only a few digital banks become enduring institutions. It took Wells Fargo 150 years; NU’s got 10.”

Warren Buffett: “It sounds like we agree NU’s done something impressive in customer experience—but the burden of proof lies in durability. The moat’s wide today, but the question is whether it gets deeper. If customers start thinking of NU the way they think of WhatsApp—‘always there, always convenient’—then you’ve got something special. But if it’s just convenience without embedded necessity, that’s fragile.”

Phase 2: Financial History & Long-term Growth

Warren Buffett: “Alright, let’s get concrete. NU’s income went from negative in 2021 to about $2 billion net income in 2024. That’s quite the swing. Free cash flow roughly doubled to $2.1 billion last year. So what do the numbers say about durability?”

David Tepper: “From a macro-financial lens, NU’s trajectory looks solid. Revenue compounded roughly 70% annually over the last five years—from $737 million in 2020 to $11.5 billion in 2024. More importantly, net margins hit 17%. Those are strong figures for an emerging market bank. Debt-to-equity at 0.45 shows conservative leverage, and ROE surpassed 33% per management’s 2025 commentary. That’s not cyclical luck; it’s scale-driven operational leverage. They turned the corner from burning capital to printing it. I see the earnings inflection as genuine.”

Dev Kantesaria: “Earnings inflection is undeniable, David, but returns on invested capital are still opaque. Their ROIC data from ROIC.AI wasn’t fully available, which makes me cautious. Rapid margin expansion often hides temporary tailwinds—credit spreads widening, inflation normalization. I need at least a decade of sustained high ROIC before I call a moat durable. NU’s history is only six years of meaningful data post-2019. That’s not sufficient proof.”

Robert Vinall: “True, but look at their internal economics through the cash flow lens. Operating cash flow moved from negative $3 billion in 2021 to +$2.4 billion in 2024—a clear improvement in capital efficiency. And they’re not returning capital yet, which is rational given reinvestment opportunities. The question for me: can they continue to compound at high returns without dilution? So far, share count expansion has been modest relative to revenue growth. That hints at disciplined capital allocation.”

Charlie Munger: “The numbers show learning capacity. NU is improving returns with scale while keeping costs lean. That’s not random. Efficiency ratio around 20%—I’ve never seen a Latin American bank operate at that level sustainably. But remember what causes failure in financial firms: mismatch between duration and liquidity. Their balance sheet is still evolving. You can’t make the mistake Lehman made—short-term financing long-term risk assets. NU must maintain moral and monetary discipline. Vélez seems aware, but human nature fights rational economics.”

Mohnish Pabrai: “And at $73 billion market cap, you’re embedding tremendous expectations. Even with $2 billion net income, that’s a 36x multiple. I’m allergic to paying 36x earnings for banks, even innovative ones. The business transition from hypergrowth to stable compounding will compress valuation multiples inevitably. When that happens, the stock could flatline for years, even if fundamentals improve.”

Pulak Prasad: “Exactly. History repeats: ING Direct once grew similarly, charming customers with simplicity. Five years later, regulation and margin pressure turned it into an ordinary retail bank. Evolutionary fragility hides behind growth numbers. NU must continuously reengineer itself to maintain cost advantage. If its AI underwriting really scales credit risk predictably, then maybe it survives. But if automation normalizes industry-wide, NU loses uniqueness.”

Warren Buffett: “Here’s what strikes me: cash is strong—$9.9 billion—but total debt’s only $3.4 billion. So they’re conservatively financed. That’s the GEICO pattern: a cost advantage amplified by financial prudence. If they can keep underwriting conservative while scaling, the long-term economics are excellent. But one bad lending cycle can erase all of that.”

Phase 3: Valuation & Final Verdicts

Warren Buffett: “Okay folks, we’ve kicked the tires. Now, at $15 a share—worth $73 billion—what do we do?”

Charlie Munger: “I like the business, Warren, not the price. You’re paying mid-30s P/E for a company in a competitive ecosystem. That’s perfection pricing. I’d be willing around $10–11 per share, where the asymmetry turns attractive. Good management, strong culture, but valuation ignores risk.”

Dev Kantesaria: “I’d avoid entirely. Fintech doesn’t meet my inevitability threshold—too many variables, too transient an edge. Even at $10, the structural certainty isn’t there.”

David Tepper: “I’d side with Charlie—buy lower. Macro and credit trends favor NU for the next few years, but it’s a cyclical trade to me, not a 30-year hold. I’d get serious south of $12, maybe $11.”

Robert Vinall: “For me, this is a hold. The reinvestment economics remain excellent. Management’s disciplined, and the franchise keeps deepening. I wouldn’t sell, but I’d wait before adding. Price needs to better reflect business maturity.”

Mohnish Pabrai: “Mathematically wrong for me. 36x P/E on a bank model—no asymmetry. I’d be out unless it reverted to single-digit multiples. Too expensive.”

Pulak Prasad: “Same here—too evolutionary a space for comfort. Even if management executes perfectly, fintech entropy guarantees constant reinvention risk. Avoid.”

Phase 4: Warren Buffett’s Synthesis & Conclusion

Warren Buffett: surveys the table, voice warm but thoughtful “Let’s try to bring this together. What struck me listening to all of you was a rare blend of admiration and caution. Everyone around this table agrees NU has executed brilliantly—turning banking frustration into delight across Latin America. That emotional connection matters. See’s Candies built its moat not on chocolate but affection. NU has that same spark with the next generation of financial customers. But affection doesn’t guarantee endurance.

Dev, Mohnish, and Pulak reminded us of that fragility—the evolutionary risk inherent in fintech. Success in this space doesn’t confer permanence the way it does in toll booths like Visa or Moody’s. Charlie and Robert see NU as disciplined, rational, increasingly efficient—a business improving with scale. David rightly argues that its balance sheet and macro tailwinds magnify those gains, but only temporarily. The long-term moat depends on whether NU’s data and AI advantage hardens into an institutional ‘must have’—something competitors can’t match.

So where does that leave us? At $15, you’re paying for perfection, not progress. The consensus leans toward admiration for the model but caution on valuation. A reasonable entry—around $10–11—would bake in margin for error while letting us participate in a potentially great compounding story. If NU proves it can lend and grow through a downturn without losing customer love, the risk-reward shifts considerably. For now, four of us would buy lower, three would stand aside entirely. That’s not a dismissal—just the discipline Charlie and I have practiced for decades: better to miss one good opportunity than catch a bad one. NU’s a fine business; the question is waiting for a fine price.”