Full Report

Industry: Latin American Retail Banking & Digital Financial Services

1. Industry in One Page

Latin American retail banking is the business of attaching consumers and small businesses to a recurring relationship — a deposit account, a credit card, a loan — and earning a spread on their balances plus fees on their transactions. The arena Nu plays in is the digital-bank slice of LatAm retail finance: same regulatory perimeter as a traditional bank (central-bank supervision, deposit protection, capital ratios), distributed through a mobile app instead of branches. The combined revenue pool across Brazil, Mexico and Colombia — interest income plus fees minus funding costs — was roughly $227.9B in 2025, growing 11% FX-neutral, per Nu's filings citing public regulator data.

The single most important thing a newcomer misunderstands: this is not a payments business. The vast majority of profits sit in credit — credit cards, personal loans, payroll-deductible loans — not in the slick app or the free transfers. The app is the distribution channel; the credit book is the earnings engine. Get that backwards and you will misread every quarter.

Consumer pays interest and fees; the digital bank captures the spread; incumbents still hold most of the pool; the central bank rewrites the plumbing every two years.

2. How This Industry Makes Money

A retail bank's revenue line has two engines: Net Interest Income (NII) — the spread between what it earns lending and what it pays for deposits — and fees & commissions (interchange on cards, fund-management fees, FX, insurance). For LatAm digital banks at scale, NII is dominant and growing: Nu's own disclosure shows NII rising from ~43% of revenue in FY2018 to roughly 79% by FY2025 as the deposit base and credit portfolio scaled. Outside analysts put credit-related interest at ~85% of Nu revenue.

Why margins exist in this industry. Three drivers, in order of importance:

1. Funding-cost gap. Customer deposits cost less than wholesale debt — often well below the policy rate (Brazil's Selic was 15% at year-end 2025). A digital bank that builds a sticky deposit franchise can lend at credit-card-level rates while paying near-zero on transactional balances. Spreads of 400–700bps are achievable in Brazil, multiples of what a U.S. or European bank earns.
2. Cost-to-serve gap. Incumbents in Brazil/Mexico/Colombia operate between 1,685 and 3,955 branches each plus tens of thousands of staff. Nu's disclosure puts its cost-to-serve and G&A per active customer ~85% below the largest incumbents (FY2025), and monthly cost-to-serve under $1.00 per active customer. That gap shows up directly in operating margin.
3. Cost of risk. Credit losses are an unavoidable expense, not a one-off. In 2025 Nu provisioned $4.2B against $11.2B of pre-provision revenue — provisioning consumed ~38% of pre-provision revenue. The winners in this industry underprice losses by 50–200bps versus peers through proprietary data and underwriting; the losers run high NPL ratios and burn equity.

Pricing units to know. Investors track three numbers, not "revenue":

• ARPAC — Monthly Average Revenue per Active Customer. Nu Q1'26: ~$16, up from $13.4 in Q3'25.
• Cost to serve per active customer — Nu Q3'25: $0.90/month. Anything below incumbent levels is the moat.
• Net Interest Margin (NIM) on interest-earning assets — the spread on the lending book.

Capital intensity is hidden, but real. This is not a software business. Every dollar of credit-card receivable consumes Tier-1 capital under Basel III. Growth in the loan book has to be funded by deposits or equity — there is no free scale.

3. Demand, Supply, and the Cycle

Demand drivers. Three forces have been pulling at the same time in LatAm — and none should be assumed permanent.

Supply constraints. Unlike a factory, supply here is regulatory capital and risk-adjusted underwriting capacity. A bank cannot grow its loan book faster than its capital + deposit base allows, and cannot grow underwriting volume faster than its credit models can absorb new vintages without losing money. New entrants need licenses — full banking licenses in Brazil and Mexico take years to secure.

Where the cycle hits first. A retail-banking downturn is a credit cycle, and it shows in this order:

Historical episodes worth remembering. Brazil's 2015–2016 recession pushed incumbent NPLs to multi-decade highs; the 2022–2023 retail-credit cycle (post-Americanas) saw credit-card NPLs in Brazil spike above 14% sector-wide before normalizing. Real losses for whoever held those receivables, not theoretical risk.

4. Competitive Structure

LatAm retail banking is a regulated oligopoly with a fast-growing digital fringe. The top five incumbents in Brazil, Mexico and Colombia hold an average of 68–80% of loans and deposits (Nu disclosure citing central-bank data). That concentration historically produced high fees, expensive credit, and weak NPS — the conditions that made disruption viable.

The four player types competing for the same wallet.

Listed-fintech peers by market cap (USD, May 2026). Nu sits between MercadoLibre and Itaú in scale; the listed-fintech tier below it (PAGS, STNE) is now sub-$3B, having compressed sharply from 2021 valuations.

Private substitutes worth tracking. Banco Inter (INTR — listed but an order of magnitude smaller than Nu), C6 Bank (private, majority-owned by JPMorgan), PicPay (private), Will Bank, and Neon. None alone threatens Nu's lead, but collectively they are the source of churn risk on the deposit and credit-card franchise.

5. Regulation, Technology, and Rules of the Game

In retail banking, regulation is not a backdrop — it is the product specification. The last five years in Brazil have seen the most aggressive rewrite of competition rules of any major banking market, deliberately engineered by the Central Bank of Brazil (BCB) to break the incumbent oligopoly.

Technology shifts that change economics, not buzzwords.

• Cloud-native cores — every digital bank runs on AWS/GCP, with marginal per-customer infra costs near zero. Incumbents are still migrating from mainframes. The 85% cost-to-serve gap is partly this.
• Real-time underwriting on proprietary data — Open Finance plus first-credit-card data creates a richer view of cash flows than bureau scores. Nu reports that ~29M Brazilians received their first credit card through it; that proprietary file is the moat.
• AI in collections and customer service — collections used to be the cost center that capped digital-bank growth; automated agents drop unit cost meaningfully.

6. The Metrics Professionals Watch

Generic banking ratios (P/E, P/B, ROE) tell you whether a bank is cheap. The metrics below tell you whether it is working.

Why these and not the usual ratios? Return on Equity in LatAm banking is often inflated by very high local rates (Brazil incumbents routinely print 18–22% ROE; Nu hit ~33% in Q4 2025). High ROE without rising ARPAC or stable NPLs is fragile — it means you're collecting Selic, not building a franchise. The unit-economic metrics above tell you whether the franchise compounds when rates eventually fall.

7. Where Nu Holdings Ltd. Fits

Nu is the scale challenger leader in LatAm digital banking — not the cheapest, not the smallest, but the only digital-first bank that has crossed into incumbent scale on customer count and is approaching incumbent scale on profitability.

Listed-peer landscape on profitability vs. scale (FY2025). Nu has the second-highest net income among LatAm-listed bank/fintech peers (after Itaú when converted to USD) on roughly the same scale of revenue base as PAGS/STNE combined.

8. What to Watch First

Five-to-seven signals that tell you whether the industry backdrop is moving Nu's way or against it. Each is observable in filings, BCB releases, or quarterly transcripts — no inside view needed.

Know the Business: Nu Holdings Ltd.

Bottom line. Nu is a credit-led Brazilian retail bank wearing a fintech valuation. The engine is a fast-growing, mostly-unsecured consumer loan book funded by an unusually cheap deposit franchise and serviced by a digital cost base that incumbents cannot match. What the market most likely underestimates is the credit-cycle exposure of a 92%-unsecured book that has yet to be tested in a Brazilian recession at this scale; what it likely overestimates is how much of today's ~30% ROE is structural versus a 15% Selic borrowed from the Brazilian Central Bank.

1. How This Business Actually Works

Nu earns money the same way every retail bank does: it pays customers near-nothing for their deposits, lends those deposits to other customers at credit-card rates, and pockets the spread. The mobile app is the distribution channel; the credit book is the engine. Net interest income now represents 79% of revenue (FY2025), up from 43% at IPO — every period that share rises, this story looks less like fintech and more like a spread bank.

Four cost advantages compound, not one. Management calls these the "four low costs" — and unlike most management framings, this one is load-bearing.

The most important number in that table is the deposit cost at 88% of CDI, not the cost to serve. Brazil's interbank rate ended 2025 near 15%; a digital bank funding itself ~12 percentage points below where a wholesale issuer would have to pay, on US$42B, is generating roughly a billion dollars of structural funding advantage before anyone makes a loan decision. The credit-card book is built on top of that funding moat, not the other way around.

A simple analogy. Think of Nu as a credit-card issuer that, instead of borrowing from capital markets at SOFR+spread, has a free checking account franchise quietly bankrolling it at half the cost of every competitor. The mobile app feeds that funnel; the credit underwriting monetizes it.

What this is not. Not a payments company, not a software business. Every dollar of credit-card receivable consumes Tier-1 capital under Basel III; growth in the loan book has to be funded by deposits or equity, regardless of how cheap the tech stack is.

2. The Playing Field

Nu sits in the middle of LatAm financial peers by scale, but the comparison set is two very different groups: incumbent Brazilian banks (lower growth, lower multiples, leveraged balance sheets) and listed fintech challengers (smaller, more volatile, mostly payments-led). Nu is the only listed name that pairs incumbent-scale ROE with fintech-scale revenue growth, and that is what justifies the multiple gap to ITUB/BBD.

Where Nu actually sits on profitability vs. growth. Itaú prints higher absolute ROE on roughly 10× balance-sheet leverage, but at 2–3% revenue growth. Nu prints almost identical ROE at less than half the leverage and ~10× the growth rate. That is the asymmetry the multiple is paying for.

What the peer set reveals. Three things, in order:

1. The relevant comparable is not ITUB; it is "ITUB in 2010." Incumbent ROEs have been propped up by the same Selic that is helping Nu — but Itaú has stopped growing customers and its loan book is mature. Nu is buying that same ROE at a much earlier point on its growth curve.
2. The fintech-listed tier has decoupled. PAGS and STNE were valued like Nu in 2021 and now trade at 1.0× book. Their problem was leading with acquiring (a margin-compressing service business) rather than credit. The cautionary tale: payments-led fintechs collapsed; spread-led fintechs (Nu, Inter) compounded.
3. MELI is the only true peer on growth-with-profitability — but MELI is a marketplace with fintech attached, not a bank. Different capital intensity, different regulatory perimeter, different exit valuation.

3. Is This Business Cyclical?

Yes — and where the cycle hits is cost of risk on the unsecured book, not revenue. Nu's credit portfolio is 92% unsecured (67% credit card + 25% unsecured personal loans, only 8% secured payroll/FGTS). That is the highest-loss-severity slice of any retail bank's balance sheet, and the company has not yet operated through a full Brazilian unemployment shock at current scale.

How the cycle shows up in this specific business.

What history says. Brazil's 2022–2023 retail-credit cycle pushed sector credit-card NPLs above 14% (post-Americanas, post-rate-hiking shock). Nu's reported NPL 90+ touched 7.0% in Q3'24 before recovering to 6.5% in Q1'26 — meaningfully better than the sector. The disclosed coverage ratios (NPL 90+ coverage 249%, Stage-3 coverage 81%) show management provisioning aggressively rather than relying on the cycle. That is the right answer — and it is also why a 30%+ ROE in the up-cycle is fragile. A 10% normalization in up-cycle ROE is not bearish; it is base case once Selic normalizes from 15%.

4. The Metrics That Actually Matter

For a credit-led digital bank, the headline numbers that move the stock (revenue growth, net income) are downstream. The five metrics below are upstream — they tell you whether the franchise is compounding or showing strain. Watch these every quarter; they will move 1–2 quarters before the P&L.

Why not just watch ROE? ROE in Brazil is partly an artifact of Selic. At Selic 15%, even a mediocre lender can print 18–20% ROE; at Selic 8%, a great lender prints 15%. ROE direction matters less than whether ARPAC is rising while cost of risk is stable — that combination tells you the unit economics are compounding independent of macro.

ARPAC dipped when Nu went mass-market (prepaid cards, NuAccount-only customers brought the average down) and has roughly tripled since 2021 as those same customers were cross-sold credit cards, personal loans, and investments. The Q1'26 print of US$16 extends the line. If this curve flattens for two consecutive quarters at constant FX, the bull case is in trouble.

5. What Is This Business Worth?

The right lens is price-to-tangible-book vs. through-cycle ROE, with growth optionality on the side — not SOTP, not EV/EBITDA, not P/FCF in isolation. A spread bank's value is the equity it carries times the return that equity earns net of the cost of equity, discounted for the risk that the spread is partly cyclical. SOTP is genuinely the wrong lens here: Nu reports as a single operating segment, has no listed subsidiaries, no holding-company discount, and ~99% of revenue runs through the same balance sheet engine.

A simple framework. At ~7.2× tangible book and ~30% ROE, Nu prices to roughly 24× through-cycle earnings if ROE normalizes to 18–20%, or roughly 15× if ROE holds at 25%+. The P/E of 28.6× on reported FY25 is misleading because reported ROE is at peak Selic. The honest underwriting question is not "what is the multiple," it is "what ROE will this business earn through a full cycle, and how much is the loan book worth growing at 20%+ for another five years."

6. What I'd Tell a Young Analyst

Five things you cannot get from a screen. In order of importance:

1. The credit cycle has not happened yet at this scale. Nu reached 100M+ customers in 2024. Its loan book grew through a benign Brazilian credit environment with falling sector NPLs. The 6.6% NPL 90+ Nu is reporting today is good — but it is also the best-case version of Nu's credit performance. The right mental discount on the ROE is "what does this look like with NPL 90+ at 8.5% and cost of risk at 50% of pre-provision revenue."

2. Read the deposit-cost line, not the customer-count line. Customer growth is a vanity metric at this scale — Brazil is saturating at 62% of adult population. Whether deposits stay at 88–91% of CDI as the base grows toward US$60–80B is the actual moat test. Incumbents will fight harder for deposits as Open Finance enables portability. If deposit cost drifts toward 95% of CDI, the funding edge halves.

3. The Mexico option is real but small until it isn't. Mexico hit break-even in Q1'26 at <1% gross-profit market share of a US$43B pool growing 15% CAGR. If Mexico ARPAC follows Brazil's curve (took 4–5 years from break-even to ARPAC > US$10), this is a 2028–2029 story, not a 2026–2027 story. Do not underwrite Mexico as if it were Brazil two years ago.

4. PIX is a feature, AI is mostly marketing. Nu's commentary about "rebuilding banking around AI" is real for collections and underwriting cost reduction, but unit economics in this business are driven by the credit decision and the funding mix, not the model. Treat AI capex as the new SaaS R&D — necessary, not differentiating.

5. What would change my mind. Three things, in order: (a) a single quarter where ARPAC rises but NPL 15–90 also rises by >50bp — revenue being bought with relaxed underwriting; the next two quarters' provisions will hurt; (b) deposit cost moves above 95% of CDI for two consecutive quarters — funding moat closing; (c) Mexico stops compounding ARPAC while still losing money — the international expansion math no longer works.

Long-Term Thesis: Nu Holdings Ltd.

1. Long-Term Thesis in One Page

The long-term thesis is that Nu becomes the dominant mass-market consumer financial services platform across Latin America by 2030–2035, compounding tangible book value at roughly 18–22% annually on three pillars: a structural funding-cost edge (sticky retail deposits priced below the interbank rate), a structural cost-to-serve edge (cloud-native infrastructure at roughly $0.80 per active customer per month versus 40–50% efficiency ratios at incumbents), and the longest reinvestment runway in Latin American banking — roughly 7% of Brazil's $100B-plus profit pool and under 1% of Mexico's $43B pool today.

This is only a long-duration compounder if through-cycle ROE clears 20% on a credit book that has never been recession-tested at $32.7B and 92% unsecured, and if the deposit franchise holds below 95% of CDI as it scales toward $80B+. The 5-to-10-year case works only if at least one of Mexico, the US, or a deepening Brazilian wallet share matures into a second meaningful profit engine before Brazil saturates. The valuation already pays for the Brazil core (28.6× trailing earnings, 7.2× tangible book against 30% ROE); the multi-year return depends on what is added to that core, not on re-rating it. The single hardest fact to internalize is that today's 30% ROE is partly rented from a 15% Selic — the durable thesis is whether the operating machine still prints 18–22% ROE at Selic 9–11%.

Thesis Strength	Durability	Reinvestment Runway	Evidence Confidence
High	Medium-High	High	Medium

---

2. The 5-to-10-Year Underwriting Map

The drivers that must work, the evidence we have today, why each could endure for a decade, and what would break the case.

The driver that matters most. Driver 1 — deposit cost as a percentage of CDI as the base scales — is the load-bearing test for the entire 10-year case. The cost-to-serve edge survives essentially any scenario; ARPAC compounding follows almost mechanically from cross-sell if the customer relationship stays primary; international optionality is priced near zero. The funding moat is what turns 18–22% ROE into a defensible compounder versus a peak that mean-reverts. If deposit cost stays below 95% of CDI on an $80B base in 2030, the operating machine is intact. If it drifts above 95% sustained, the bull case requires a different argument.

---

3. Compounding Path

The long-term path is a book-value compounding story, not a revenue story. With ROE at 30% in 2025 and capital allocation 100% reinvested (no dividend, minimal buyback, $7M of FY25 capex), the returns calculus reduces to: at what through-cycle ROE does the book compound, and for how long can it compound before mean reversion?

Tangible book per share roughly doubled from FY2023 to FY2025 ($1.18 to $2.09) as the loss-making investment phase ended and retained earnings compounded into book. With ROE at 30.3% and dilution under 1.5% per year, the math is mechanical: book per share grows at ROE minus dilution minus the (currently zero) payout ratio, which is roughly 28–29% per year today.

The arithmetic. At 20% through-cycle ROE and no payout, tangible book per share roughly doubles every four years. From $2.09 today, base case lands at $5.20 by 2030 and $11.50 by 2035. At a P/TBV of 4× on the 2030 number, that implies roughly $20 per share — about 65% above the May 2026 price of $12.18. The bull case requires either a higher ROE (25%+) for longer, or a higher exit multiple sustained by international optionality earning its capital. The bear case is not that book stops growing but that the multiple compresses faster than book compounds — a 3× P/TBV on $3.60 of 2030 book is below today's price.

Three observations on the cash side. OCF/NI has held at 1.22× for three consecutive years — earnings convert to cash above 100%, which is structurally what a deposit-funded compounder should do. Capex is essentially zero ($7M FY25 in the cash flow statement; $341M in MD&A capex that includes capitalized software, still under 4% of revenue). The capital intensity of growth is not the OCF/NI line; it is the Tier-1 capital constraint of growing an unsecured loan book — and at 6.6× assets/equity versus 13–14× at incumbents, Nu has roughly 2× the balance-sheet headroom before leverage starts diluting ROE.

---

4. Durability and Moat Tests

Five tests — three competitive and two financial — that determine whether the moat survives the next decade.

Reading the moat trajectory. The cost-to-serve advantage strengthens in absolute terms as scale grows but narrows versus the digital-peer cluster (Inter, C6) that runs on similar rails — net, the lever drifts from a 5 to a 4. The deposit-funding edge is the most vulnerable lever over a decade because a Selic normalization mechanically halves the absolute funding advantage even if percentage holds, and Open Finance enables structural deposit portability — most likely lever to step from 5 to 3. The credit-data, engagement, and scale-economy levers strengthen as cohorts mature and AI capex compounds. The contractual switching-cost lever is regulatorily suppressed and will not improve. Net: the moat broadens in some directions and narrows in others; the durability question is whether the structural cost edge and behavioral engagement are enough to substitute for a compressing funding gap.

---

5. Management and Capital Allocation Over a Cycle

Founder David Vélez has been CEO for 13 years through the IPO, two LatAm launches, the 2022–23 credit cycle, the profitability inflection, and now the AI/US pivot. His personal $11B stake (74.4% of votes on 18.6% of economics) is the largest absolute founder skin-in-the-game of any newly-listed bank globally, and his voluntary 2022 termination of a $423M Contingent Share Award — to save shareholders up to 2% of dilution — remains the single most concrete piece of alignment evidence in the file. CFO Guilherme do Lago has held the seat through the entire profitable phase; CEO Brazil Livia Chanes (the 80% of revenue) has operated through the 2024 NPL peak and recovery. Roberto Campos Neto's mid-2025 appointment as Executive Vice-Chairman has strong operating logic (regulatory dialogue, US bank build-out) but represents a textbook revolving-door optic for a bank whose payment-license framework he supervised seven months earlier.

The capital allocation track record is consistent with the long-term thesis: net income redeployed 100% into the loan book, Mexico/Colombia expansion, and now the US bank, with essentially no dividend, no material buyback ($6.2M FY25), and no M&A bloat (Hyperplane $1.5M FY25 and Tyme $250M minority stake are the largest disclosed transactions in three years). At 30% reported ROE and 6.6× leverage versus 13–14× at incumbents, the math says retaining all earnings is the right answer until either the international franchise is meaningfully profitable or Brazilian saturation closes the reinvestment runway — neither has happened yet.

The 10-year capital-allocation question is when management initiates a payout. At current pace, Brazil ARPAC compounding plus Mexico ramp should consume reinvestment capacity through 2028–2029; after that, a dividend or buyback program becomes the natural next step. Whether management initiates capital-return discipline before book per share compounds past a level where reinvesting at 20%+ ROE is no longer plausible is the test. Two precedents to watch: ITUB and BBD both initiated heavy dividend programs in their early-2000s maturation; Nu has not yet shown what its capital-return cadence looks like, and the disclosure machinery (founder-controlled FPI, no individual NEO comp, no Say-on-Pay) is weaker than it would be at a US domestic.

The bear-side governance critique is real but not load-bearing for the 10-year case: insider activity since IPO has been sell-only (Vélez August 2025 $462M block, Junqueira March 2026 $4M open-market sale, Sequoia 13G/A cut 57% in May 2025); the introduction of the Managerial P&L framework in Q4 2025 with ISAE 3000 limited assurance only reduces backward comparability; ~$243M of FY25 development spend was capitalized rather than expensed. None of this is fraud — Nu sits in the forensic "Watch" band, not the elevated band — but the machinery for catching aggressive judgment calls runs through fewer hands than at a US domestic, which argues for a margin-of-safety adjustment on the multiple rather than a thesis breaker.

---

6. Failure Modes

Six specific ways the long-term thesis breaks, each with observable early warnings rather than generic "execution risk."

---

7. What To Watch Over Years, Not Just Quarters

Five observable multi-year milestones, each with its time horizon and the validating/refuting reading.

Competitive Position: Nu Holdings Ltd.

Competitive Bottom Line

Nu has a real, measurable moat in cost-to-serve and funding cost — not slogans, numbers any peer can be benchmarked against. The cleanest test: Nu generates roughly the same return on equity as Itaú Unibanco on less than half the balance-sheet leverage while still growing revenue at 27% versus Itaú's 3%. The one competitor that matters most for the next 24 months is not Itaú — incumbents are losing customer wallet, not defending it — it is Mercado Pago (MELI), the only regional rival with a comparable cost structure, a self-funding merchant ecosystem, and an accelerating fintech book. The long-tail risk is the private digital cluster (Banco Inter, C6 Bank, PicPay, Banco BV) chipping at deposits as Open Finance enables portability.

The Right Peer Set

Five primary peers split into two groups: incumbent universal banks (Itaú, Bradesco, Santander Brasil) that own the profit pool today, and listed fintech rivals (MercadoLibre, StoneCo) that share Nu's distribution model. PagSeguro is included as a supplementary peer — closer in operating model to Nu than the banks but smaller in scale.

Why these five (plus PAGS) and not others. Banco Inter (NASDAQ: INTR) is the natural sixth — the most-cited Brazilian digital-bank rival in industry and Warren research — but Fiscal.ai 404'd on the listing ID; it is treated qualitatively in the Threat Map. State-owned banks (Banco do Brasil, Caixa) are excluded because their cost of capital is set by policy not the market. BTG Pactual and XP target a different customer segment (wealth/affluent) than Nu's mass market. Global neobanks (Chime, SoFi, Wise, Revolut) operate in different regulatory regimes; useful for sentiment, not comparison.

Growth vs profitability bubble. Nu and MELI are the only LatAm-listed digital-finance names jointly clearing 25% revenue growth and 25%+ ROE. Itaú is a profitable cash machine that has stopped growing customers. Bradesco and Santander Brasil are leveraged spread banks with mid-teens ROE at best. STNE and PAGS — the previous fintech vintage — shrank because they led with acquiring (a margin-compressing service) instead of credit.

Where The Company Wins

Four concrete advantages, ranked by how hard each one would be for a competitor to copy.

1. Funding cost — 88% of CDI on $42B of deposits

This is the load-bearing edge. With Brazil's Selic at 15% at year-end 2025, Nu paid roughly 13.2% on its deposit base versus the 15% wholesale rate every competitor without a sticky retail deposit franchise pays. On $42B of deposits, that 180bps gap is about $760M of annualized funding advantage before anyone makes a loan decision — and the cheapest in the LatAm-listed digital-finance set. Mercado Pago depends largely on payment-account float that is structurally smaller per user; PagBank and StoneCo run far smaller deposit bases relative to credit assets.

Why competitors cannot copy this quickly: it requires being a customer's primary banking relationship, which in turn requires sustained app engagement, salary/cash-in flows, and trust. Nu reports 90% of Brazil customers hold balances and an 83% monthly activity rate — engagement closer to a social network than a bank. (Source: Nu Q1'26 release, Warren tab.)

2. Cost-to-serve — $0.80/month per active customer; efficiency ratio 17.6%

Nu's cost-to-serve is roughly 85% below the largest Brazilian incumbents, per the company's disclosure and consistent with industry research. The efficiency ratio (operating expenses as a % of revenue) was 17.6% in Q1'26, compared with operating expense ratios in the 40–50% range typical of Brazilian incumbents. That is the structural payoff of cloud-native infrastructure, no branches, and 14,314 customers per employee versus ~1,234 at the incumbents.

3. Same ROE as the incumbent leader on half the leverage

The cleanest single comparison: Itaú prints a 21% ROE on 14.3× assets-to-equity leverage. Nu prints 30% on 6.6× leverage. Nu's return on assets — the underwriting and pricing skill — is roughly 2.7× Itaú's (4.5% vs ~1.6% ROA equivalent). When Selic normalizes from 15% toward 9–11%, the gap will narrow, but the leverage asymmetry says Nu has room to either grow the book or take on more leverage without diluting ROE — incumbents do not.

4. Customer scale and engagement — at the same level as the largest incumbents

Nu reached 135M customers globally in Q1'26, approaching 100M active customers in Brazil alone, with a stated 62% adult-population penetration in Brazil. That is at parity with — and in some segments ahead of — Itaú and Bradesco by customer count. The Q3'25 release cites Brazil's Central Bank ranking Nu as the third-largest financial institution in Brazil by number of customers. In Mexico, Nu reached 15M customers and the third-largest position by customer count as of Q1'26. Incumbents took decades and thousands of branches to build comparable bases.

Where Competitors Are Better

Four concrete weaknesses where a peer is clearly stronger today.

1. Itaú still earns ~6× more dollar profit than Nu

Itaú's FY2025 net income was $16.6B USD-equivalent, versus Nu's $2.87B — a 5.8× gap. Even after a decade of customer-share losses to challengers, Itaú monetizes its existing relationships more deeply: corporate banking, investment banking, asset management, insurance, and wealth franchises Nu has not built. Nu has won the consumer onboarding race; it has not won the profit-per-customer race, where incumbents still extract more revenue from a smaller but wealthier book. The Itaú threat is not in net adds — it is in defending the affluent and high-margin slices.

2. MELI grows faster, with a self-funding merchant flywheel

MercadoLibre grew revenue 39% YoY in FY2025 vs Nu's 27%, with a 36% ROE — higher on both axes. Mercado Pago has something Nu does not: captive merchant payment flow from the Mercado Libre marketplace, which both pre-funds its credit book (Mercado Crédito) and acquires new fintech users at near-zero cost. Nu's acquisition is paid plus word-of-mouth. If consumer-finance and merchant-acquiring continue to converge, MELI's structural advantage in merchant relationships becomes a real competitive limit on Nu's SMB ambitions.

3. Incumbents own the corporate, SME, and wealth franchises Nu has not built

Itaú, Bradesco, and Santander run mature corporate banking, investment banking, payroll-deductible lending (consignado), and wealth-management businesses generating high-margin, low-volatility fee income. Nu has none at scale. Bradesco alone reported BRL 47.6B of non-interest income in FY2025 (about $8.7B USD-equivalent) — over 3× Nu's entire fee-income line. Not a weakness Nu needs to fix tomorrow, but it caps revenue diversification and explains why incumbents look durable even with sub-15% revenue growth.

4. SMB acquiring — StoneCo and PagSeguro are deeper

StoneCo had 4.8M SMB clients as of December 2025, and PagSeguro reports a roughly comparable SMB acquirer base; both have years of merchant data, sales feet on the street, and POS-device deployments Nu would have to build or buy. Nu's SMB push (Nu Negócios) is real but sub-scale. If the next leg of LatAm fintech is SMB-as-credit-customer rather than consumer-as-credit-customer, STNE/PAGS are better positioned at the acquiring touchpoint despite having lost the consumer game.

Threat Map

The useful frame is not "who will displace Nu" — nobody is close — but "who chips at which part of the business model first, and by when."

Moat Watchpoints

Five signals — observable in filings, BCB releases, or competitor disclosures — that tell you whether Nu's competitive position is improving or weakening. Each is leading rather than lagging.

Current Setup & Catalysts: Nu Holdings Ltd.

1. Current Setup in One Page

The stock trades at $12.18 on 15 May 2026, in the bottom 4% of its 52-week range and ~35% below the $18.83 January peak, because the market has reframed Q1 2026 from a record revenue print into the first real-time stress test of the unsecured credit book.

The setup is mixed-to-bearish: revenue (+57% YoY), Mexico break-even, and ROE at 29% all came in record-strong, but risk-adjusted NIM compressed 100 bps QoQ to 9.5%, 15–90 day NPL jumped 89 bps to 5.0%, and management labelled the move "intentional risk expansion" via the NuFormer credit model — language the tape priced as the leading edge of a credit cycle. The 5-to-10-year thesis hinges on through-cycle ROE clearing 20% and the deposit funding moat holding below 95% of CDI; neither has been refuted, but both are now testable rather than theoretical.

The single most consequential near-term event is the Q2 2026 print in mid-August 2026 because it is the first quarter that decides whether Q1's NIM/NPL move was seasonality (management view) or trend (market view). The next six months also contain the October 2026 Brazilian general election, the BCB Selic path on a 15% policy rate, and the FY2025 20-F filing with the first comparable-period managerial P&L re-presentation.

Recent setup rating: Mixed.

---

2. What Changed in the Last 3-6 Months

The setup has been rewritten almost entirely in the last six months. Three things shifted: the credit-quality leading indicator turned for the first time since 2024, the international optionality became real (Mexico break-even + OCC approval), and the chart broke even as the fundamentals held records.

The narrative arc since November. For 18 months before November 2025, the question was whether Nu could keep printing 30%+ ROE while expanding into Mexico and pivoting toward AI. Then four things happened in rapid sequence: the FGTS rule changed (1 Nov), Brazil signalled mandate-expansion risk (mid-Dec), the company introduced a non-IFRS managerial framework as the primary headline (26 Feb), and Q1 2026 showed the first NPL up-tick + NIM compression since 2024 (14 May). The market did not change the long-term thesis; it raised the disconfirming-evidence bar. The unresolved question is whether Q1's NIM/NPL move was seasonality (NuFormer-driven "intentional risk expansion" management says is decoupled from cycle) or the leading edge of the credit cycle bears have warned about for two years.

---

3. What the Market Is Watching Now

---

4. Ranked Catalyst Timeline

Ranked by decision value to an institutional underwriter, not by chronology. The top three each update a different load-bearing thesis variable.

---

5. Impact Matrix

Five items that actually resolve the debate, rather than merely add information.

The pattern: three of the five items resolve in the next six months (Q2 print, election, tax rate). Two are multi-year (deposit cost, US bank operationalization) but each has a near-term observable. No thesis-level catalyst requires waiting beyond 90 days.

---

6. Next 90 Days

---

7. What Would Change the View

Three observable signals would force a real underwriting update over the next six months.

First, Q2 2026 NPL 15–90 below 4.7% with risk-adj NIM holding at or above 9.5% would invalidate the bear's credit-cycle read and re-validate the through-cycle ROE pillar — consistent with the sell-side $19 consensus and an end to the death-cross regime.

Second, the opposite — two consecutive quarters (Q2 + Q3 2026) with NPL 15–90 at or above 5.0%, NPL 90+ coverage falling below 220%, and deposit cost crossing 92% of CDI — breaks two of the three load-bearing thesis variables simultaneously (through-cycle ROE on the unsecured book and the deposit funding moat) and is consistent with the bear's $8 downside scenario.

Third, the Brazilian October election outcome plus any new consumer-credit caps (interchange, personal-loan rates, interest-on-equity restrictions) is the single regulatory event with power to compress the highest-yielding piece of the book; absent that, the multiple is mostly resolved by what the credit data and deposit franchise print.

None of these are next-quarter noise; each updates a long-term thesis variable named in the durability tests and failure modes of the Long-Term Thesis tab. The tape has front-loaded the bear read; the burden of proof now sits with Q2 numbers that refute it.

Bull and Bear

Verdict: Lean Long, Wait For Confirmation — the unit economics gap (4.58% implied ROA vs Itaú's 1.47%) is real and mechanical, but Q1'26's simultaneous NIM compression (10.5% → 9.5%) and NPL drift (4.1% → 5.0%) make the load-bearing variable — through-cycle ROE — genuinely contested for the first time.

Bull wins on the underlying machine: 30.3% ROE on 6.6× leverage with an 88%-of-CDI funding base is not a story you can construct without a real structural edge; Mexico break-even plus the OCC charter sit as unpaid optionality in the multiple. Bear wins on the question that decides the multiple: whether 25%+ ROE survives a Selic cut from 15% to 9–11% on a 92%-unsecured book that has never been recession-tested at $32.7B. The tension that resolves the debate is whether the Q1'26 NPL/NIM step-down is seasonal mix or the leading edge of normalization. Two prints — Q2'26 (August) and Q3'26 (November) — produce the evidence; until then, conviction belongs to whichever side prints the next quarterly data point.

Bull Case

Bull's scenario value: ~$22 over 18 months (through end-2027). Method: ~18× FY2027E EPS of ~$1.20 (consensus FY26 EPS $0.89 growing ~35%), triangulated against ~6.5× tangible book of ~$3.40 (TBVPS $2.09 today compounding ~25%/yr at 28%+ ROE). Both methods converge to ~$21–$23. What would confirm: Q2'26 and Q3'26 prints showing NIM stabilization at or above 9.5% with NPL 15–90 flat or improving from the 5.0% Q1'26 level. Disconfirming signal: deposit cost above 95% of CDI for two consecutive quarters — the funding-moat pillar — forces re-underwriting of the entire thesis.

Bear Case

Bear's downside scenario: ~$8/share (≈34% below $12.18, ≈60% from the Jan'26 peak of $18.83) over 12–15 months. Method: P/TBV compression from 5.8× to ~3.8× on roughly flat tangible book of ~$2.10/share (cycle ECL absorbs retained earnings), equivalent to ~13× through-cycle EPS of $0.60 (ROE normalized to 17–19% on $11–12B equity). Primary trigger: a Q2 or Q3'26 print where NPL 15–90 stays at or above 5% and NPL 90+ coverage falls below 220% and deposit cost crosses above 92% of CDI for two consecutive quarters — that triple breaks both load-bearing pillars (recession-tested credit edge and deposit stickiness). Refutation signal: a single quarter where NPL 15–90 prints under 4.5%, coverage above 240%, deposit cost under 90% of CDI on a $45B+ deposit base, and Mexico ARPAC clears $10 with positive net income.

The Real Debate

Verdict

Lean Long, Wait For Confirmation. Bull carries more weight because the underlying machine is observable, mechanical, and structurally rare — a 4.58% implied ROA on a 17.6% efficiency ratio against incumbents stuck at 1.06–1.47% ROA is not a number you fabricate, and deposits have held in an 78–91%-of-CDI band through a 2.2× scaling of the base.

The single most important tension is whether Q1'26's simultaneous NIM compression and NPL drift is seasonal mix or the first quarter of a credit cycle on a $32.7B, 92%-unsecured book that has never been tested at scale. That answer determines whether through-cycle ROE clears 25% (bull right) or settles at 18–20% (bear right, multiple compresses to 4–5× book).

Bear could still be right: Brazilian unsecured-credit cycles have historically been steep, the FGTS regulatory cliff just hit, the October 2026 election adds tail risk, and the governance machinery (founder-controlled FPI, ISAE 3000-only Managerial P&L, $243M of capitalized intangibles, insider selling) is exactly what runs through fewer hands when judgment calls expand.

The durable thesis breaker is deposit cost crossing 92% of CDI for two consecutive quarters, which would prove the funding moat is regime-dependent; the near-term evidence marker is NPL 15–90 in the Q2'26 print (August), which arbitrates the credit-cycle question without yet deciding the multiple. Until one of those data points lands, the institutional posture is to hold the thesis in a watchlist with intent to engage on Q2'26 evidence, not to size against a hostile tape with the load-bearing variable contested. The unit-economics edge is too clean to dismiss and the optionality (Mexico break-even, OCC charter, Campos Neto) is priced at zero — but two earnings prints stand between today and conviction.

Moat: Nu Holdings Ltd.

1. Moat in One Page

Verdict: narrow moat. Nu has two genuine, measurable, company-specific economic advantages — a digital cost-to-serve roughly 85% below the largest Brazilian incumbents, and a primary-relationship deposit franchise that funds itself at 88% of Brazil's interbank rate (CDI) on roughly $42B of balances. Those facts show up in the numbers that matter: a 30%+ ROE earned on roughly half the leverage of Itaú, an efficiency ratio of 17.6% in Q1'26 versus 40–50% at incumbents, and 17M FY2025 net adds at a cost-to-acquire of about $7.40.

It is not yet a wide moat. The book has never been tested in a Brazilian recession at current scale, a meaningful piece of today's 30% ROE is borrowed from a 15% Selic, Open Finance and PIX are explicitly engineered to lower switching costs across the system, and the cost-to-serve advantage exists primarily versus incumbents — not versus the digital cluster (Banco Inter, C6, PicPay, Mercado Pago) on similar cloud-native rails. Geographic concentration (~85% of revenue in Brazil) and founder-controlled governance limit external checks.

Moat rating: Narrow. Weakest link: funding edge partly borrowed from 15% Selic.

---

2. Sources of Advantage

The candidate moat sources, with evidence and proof quality. Quick definitions: switching costs = friction (data, paperwork, lost benefits, retraining) a customer faces moving to a competitor. Network effects = value of the product to one customer rises as more customers join. Cost advantage / scale economies = lower unit cost than competitors, from structural choices a rival cannot replicate quickly. Intangible assets = brand, data, license, or trust that competitors cannot buy.

Reader's note on weights. Two of the nine sources do the load-bearing work — cost-to-serve and deposit funding. Brand/trust and credit data are real but secondary. Switching costs are not a moat for this name; the regulator has built rails to make them weaker. Anyone arguing for a wide moat is implicitly betting that the cost edge survives challenger imitation and that behavioral stickiness in primary deposits substitutes for missing contractual switching costs.

---

3. Evidence the Moat Works

The cleanest tests of whether the alleged moat shows up in the actual income statement.

The headline number. Implied ROA from peer disclosures separates Nu from incumbents and from challengers.

Nu's implied ROA of 4.58% sits at roughly 3× the Brazilian incumbent average and beats every other listed LatAm consumer-finance peer except MELI (a marketplace-funded fintech, not a stand-alone bank). That is the math the moat is doing.

---

4. Where the Moat Is Weak or Unproven

The honest places to push back on the bull case.

---

5. Moat vs Competitors

The useful frame is not "who is Nu's biggest threat" — incumbents are losing wallet, not defending it — but which competitor is best positioned on each moat lever.

Where Nu actually sits. Implied ROA (clean read of cost + underwriting net of leverage choice) versus revenue growth (operational test of compounding). Nu and MELI are the only two names in the upper-right quadrant; incumbents are top-left (slow + profitable on leverage); STNE/PAGS in the middle (challengers that picked the wrong leg of fintech).

---

6. Durability Under Stress

A moat only matters if it survives stress. The seven stress cases below are the ones a portfolio manager should pre-rehearse for this name.

Where this leaves the durability call. Cost-to-serve and engagement levers should survive a normal cycle — structural, not policy-driven. The funding edge is most exposed to Selic normalization and Open Finance maturation; it carries the largest near-term downside. The credit-data edge needs to pass a recession test before it can be underwritten as a moat rather than a hypothesis.

---

7. Where Nu Holdings Ltd. Fits

The moat does not live evenly across the business. It is concentrated in one geography, one customer segment, and two products — and the company is open about which lines are still building.

One protected segment, one commodity segment. The Brazil mass-market consumer credit-card + primary-deposit relationship is the protected segment — what 7× tangible book and 30% ROE are paying for. SMB banking, wealth management, Colombia, and (so far) the US franchise are commodity segments where Nu has no proven advantage and is still building one. The investment question is whether the protected segment can sustain a 5%+ implied ROA for long enough to make the commodity segments matter.

---

8. What to Watch

The leading-not-lagging signals to monitor. Each is observable in quarterly filings or BCB releases.

Financial Shenanigans

The Forensic Verdict

Nu Holdings sits in the Watch band on the forensic risk scale. Clean cash-conversion record, effective internal-control attestation, and improving credit metrics on one side; founder-controlled dual-class governance, opaquely disclosed compensation budget, in-flight "Managerial P&L" framework rollout, and visible intangible-asset capitalization on the other. Nothing in the public record points to revenue manipulation, restatement, or auditor concern. The two findings worth underwriting: (1) governance concentration that limits investor checks on judgment-laden accounting choices and (2) the Q4 2025 disclosure-framework reset — both raise the cost of trusting reported numbers without external corroboration. The single data point that would most change the grade is any future material weakness, qualification, or KPI definition change that breaks comparability with FY2024 financials.

Shenanigans scorecard — all 13 categories

Breeding Ground

The conditions that make accounting shenanigans more likely are present, but in a benign rather than predatory configuration. Concentration of power is the dominant signal: founder David Vélez Osorno is simultaneously CEO and Chairman, controls 88.3% of Class B shares (20 votes per share) for 74.3% of voting power, and the Shareholder's Agreement gives him a veto over related-party transactions, indebtedness above net equity, and other material actions. Nu is a "controlled company" under NYSE rules and is exempt from the majority-independent-board and fully-independent-comp-committee requirements. As a Cayman Islands foreign private issuer, individual NEO compensation is not disclosed; investors only see the aggregate $91.3M figure for FY2025 covering all directors and key management personnel. Form 4 insider filings are not required for FPI directors and officers.

The offsetting governance evidence is meaningful: an independent Lead Director (Anita Sands), an Audit & Risk Committee chaired by Rogério Calderón — a former PwC Brazil audit partner with a decade in that role — and seven of nine directors classified independent. The board added Roberto Campos Neto (former Banco Central do Brasil President) and CTO Eric Young in 2025. A compensation clawback policy was adopted in October 2023, and the FY2025 filing states no restatement-triggered recovery has been required.

The structure is not predatory — Vélez is a long-tenured founder with reputational skin in the game and a Sequoia-led board with operating history — but the machinery for catching aggressive accounting choices runs through fewer hands than it would at a U.S. domestic, single-class, fully-independent board. That is the right frame for breeding-ground risk on this name.

Earnings Quality

Reported earnings look earned, not manufactured — the qualifications are about classification, not creation. Net income reconciles cleanly to the cash flow statement, revenue is overwhelmingly interest income (79% of total in FY25) and interchange fees (recognized at point of transaction), and there is no material one-time-gain reliance. Revenue growth is decelerating naturally with scale: +144% (FY22) → +85% (FY23) → +45% (FY24) → +29% (FY25). Operating leverage is real: total non-interest expense as a share of revenue dropped from 67% (FY22) to 28% (FY25).

The one judgment-zone test that should make a forensic reader pause is the balance-sheet treatment of operating costs. Per the FY2025 MD&A, capital expenditures (cash payments for PP&E or intangible assets) were $340.8M in FY25 versus $175.0M in FY24 — nearly doubling. The cash-flow statement's "capex" line of $7.2M captures only property, plant and equipment; the difference is software, customer-onboarding, and product-development costs that get capitalized as intangibles. Intangibles on the balance sheet grew from $295.9M (FY23) to $601.7M (FY25), a 103% increase against 87% revenue growth. Depreciation and amortization is $98M FY25 — well below the gross capitalization pace.

Practical read: software and product-development costs that other digital banks might expense are sitting on the balance sheet and amortizing slowly. None of this is improper under IFRS, but the gap between cash capex ($340.8M) and D&A ($98M) means roughly $243M of FY25 development spend was capitalized rather than expensed — a benefit to current-period operating profit of similar order. If growth slows or product launches are written down, this becomes a future impairment risk.

Reserve and provisioning behavior is the other forensic-quality question. The company applies an expected-credit-loss (ECL) model that front-loads provisioning at loan origination. ECL as a share of total receivables held steady at 15.4% across FY24 and FY25, even as the loan book grew 58%. NPL 90+ in Brazil improved from 7.0% (FY24) to 6.6% (FY25); NPL 15-90 improved from 4.1% to 3.9%. Provisions grew +33% against +58% portfolio growth — the model's "front-loading" claim is doing some of the lifting here, but the credit data trend supports the lower marginal provisioning rate.

The effective tax rate trajectory is the third earnings-quality item to track. ETR fell from 33.0% (FY23) to 29.4% (FY24) to 25.8% (FY25), versus a Brazilian statutory rate of 40%. Q1 2026 came in at 8.7% — a $82.9M tax expense on $954.3M of pretax income. That single-quarter rate is almost certainly driven by Brazilian "juros sobre capital próprio" (interest on equity) deductions or deferred-tax-asset recognition rather than ongoing tax mix; it is not a sustainable run-rate. Investors valuing forward EPS off Q1 2026 risk anchoring on a transitory tax benefit.

Cash Flow Quality

CFO is durable but is essentially a deposit-funded operating book, and that single fact governs how to read the next four years of cash conversion. Three-year aggregate CFO of $7.17B versus net income of $5.87B gives a CFO/NI ratio of 1.22 — clean by industrial standards but mechanically driven by the bank business model: deposit growth shows up as an operating cash inflow, and loan growth shows up as an operating cash outflow. Both are large. In FY25 alone, deposits grew $13.0B, the loan book grew $4.1B, and credit card balances grew $7.1B.

The "free cash flow" line on the cash flow statement ($3,493M in FY25) is misleading on its own because it subtracts only PP&E capex of $7.2M, ignoring the $340M of intangible-asset additions that the MD&A includes in its capital expenditure definition. Re-cast FCF using the MD&A capex definition gives roughly $3,160M for FY25 — still strong, but $333M lower than the headline.

The accrual ratio is negative at -1.0% of average assets in FY25 — net income is smaller than CFO by roughly 1% of the asset base, which is the opposite signal from a typical earnings-management concern. Working-capital quality is also clean in the sense that there is no factoring, no off-balance-sheet receivables sale, and no supplier finance program disclosed. The cash flow statement reconciliation discloses ECL of $4,701.1M as an add-back (versus $4,204.9M on the income statement) — the $496M difference reflects FX translation and other movements through the credit-loss line, not a hidden expense.

Acquisitions are immaterial: $1.5M in FY25 and $5.6M in FY24. Hyperplane (Jul 2024) and the Tyme minority stake (Dec 2024) do not move the cash flow story. Goodwill held flat at roughly $410M across FY23-FY25, which is what you want from a non-acquisitive compounder — no purchase-accounting tailwind to revenue or working capital.

Metric Hygiene

The metric-hygiene risks are the most concrete forensic issues for this name, and they cluster around three items.

The "Managerial P&L" change is the most consequential single disclosure event of FY2025 from a forensic perspective. Per the FY2025 20-F, "in the beginning of the fourth quarter of 2025, we introduced the Managerial P&L, representing an evolution in our disclosure framework… a structural, complementary reorganization of line items designed to enhance comparability as the business scales." Management states the change "preserves net income, cash flow, and capital." A separate Q1 2026 6-K confirms an independent firm provided limited assurance under ISAE 3000 over the Managerial P&L reconciliation report — not a full audit. Limited assurance is appropriate for management-defined metrics, but investors should not equate it with audited financial statements.

The persistent $80-135M gap between SBC as reported in the cash flow statement and SBC added back in the Adjusted Net Income reconciliation deserves a diligence question to investor relations. It does not appear to be improper — the gap likely reflects different scopes (e.g., inclusion of social-charge accruals or different vesting treatment) — but the company's filings do not reconcile the two numbers, and a reader cannot derive the difference from the extracted disclosures.

What to Underwrite Next

The forensic risk here is a margin-of-safety adjustment, not a thesis breaker. The accounting is conservative on most tests, governance is concentrated but not yet litigious, and there are no smoking guns. Position sizing should respect the limits of disclosure; valuation should not anchor on Adjusted NI or Q1 2026 tax-rate normalization.

The five items to track next quarter and into the FY2026 20-F:

1. Managerial P&L re-presentation. Confirm that FY2024 and FY2023 financials are re-presented to the new framework in the FY2026 20-F so analysts can build like-for-like trend lines below revenue. If only forward periods are presented, comparability is broken.
2. Effective tax rate normalization. Watch FY2026 quarterly tax expense relative to pretax. A run-rate of 22-27% is plausible given Brazilian interest-on-equity deductions and international mix; a sustained sub-15% rate would imply a structural shift that needs explanation in the next 20-F notes.
3. Intangible-asset capitalization vs amortization. Track quarterly intangible additions, D&A, and any impairment charges. The MD&A capex figure ($340.8M FY25) should grow no faster than revenue; sustained outpacing implies a thicker future amortization wedge or impairment risk.
4. ECL coverage if loan growth slows. With the model front-loading provisioning at origination, a sharp deceleration in new originations (the FGTS regulatory cliff from November 2025 is a near-term test) should mechanically reduce provisioning expense even if late-stage NPLs deteriorate. Watch the NPL 90+ trend and Stage 3 coverage in the FY2026 audited financials.
5. Any change to non-IFRS definitions. Adjusted NI currently adds back SBC after tax. Adding new line items (e.g., excluding "transformation costs", "AI investment", or international-expansion losses) is a classic earnings-management pattern. Watch the FY2026 reconciliation for new add-backs.

The grade would downgrade to Elevated (41-60) if: (a) FY2024/2023 financials are not re-presented on the Managerial P&L framework, breaking comparability; (b) any new non-IFRS add-back is introduced that excludes recurring costs; (c) auditor changes or any material weakness is disclosed; (d) intangibles growth continues to outpace revenue by 15+ points; or (e) ECL coverage as a share of receivables declines while NPL 90+ moves higher.

The grade would upgrade to Clean (0-20) if: (a) the Managerial P&L is reconciled to fully-audited historical financials for three prior years; (b) individual NEO compensation is voluntarily disclosed despite the FPI exemption; (c) the SBC gap between cash flow statement and Adjusted NI is reconciled in plain English; and (d) ECL coverage and NPL 90+ continue their FY2025 improvement.

Bottom line. The accounting risk on Nu Holdings is the kind a long-only manager should respect through position sizing — call it a 100-200 bps haircut to the multiple investors would otherwise pay for a digital-bank with this growth and ROE. It is not a short thesis on its own, and there is no evidence in the public record of revenue manipulation, classification gaming for headline earnings, or auditor concern. The risk is concentrated governance, not concentrated fraud — and the difference matters.

The People

Governance grade: B–. A founder with $11B of personal Nu stock and a credible bench (ex–Brazil central-bank chief, ex–Square comp chair, ex–PayPal president), offset by a hard-coded super-voting structure that gives David Vélez 74.4% of the vote on roughly 19% of the economics — and lets him veto major capital actions even if his stake falls to 5%.

1. The People Running This Company

Nine senior executives. The decisions that move the P&L sit with three of them: Vélez (strategy/capital), do Lago (CFO and the credibility relay to the market), and Chanes (the Brazil P&L, ~80% of revenue). The new external hires — Campos Neto, Young, Eismann — are the genuine swing variables on whether Nubank can keep operating-leverage gains while building a global stack.

2. What They Get Paid

Nu discloses aggregate compensation only — a Foreign Private Issuer exemption Vélez has used since IPO. No individual named-executive-officer numbers, no Form 4 filings, no Say-on-Pay vote. For FY2025 the company says aggregate compensation of all directors and key management was US$91 million.

Is the pay sensible? At $11.2B revenue and $2.87B net income, $91M of aggregate comp is 0.8% of revenue and 3.2% of profit — well below US bank-fintech peers. The 80%-plus stock weighting (per Nu's own 2022 clarification of the Contingent Share Award structure) means most of the headline number is at-risk equity that pays out only on hitting CSA performance hurdles. The honest critique is not that pay is excessive; it is that outside shareholders cannot see who gets what, so they cannot verify whether the CFO/CRO/Brazil-CEO are paid commensurate with the value they create.

3. Are They Aligned?

This is the section that decides the file. The arithmetic of alignment at Nu is unusually asymmetric.

Control versus alignment

Class B shares carry 20 votes per share and are not publicly traded. Vélez holds 88.3% of Class B through Rua California Ltd. (which he owns 100% of) plus a small spousal block. Junqueira holds another 11.7% of Class B via trusts. Combined founder voting power: ~84%. Outside Class A shareholders — collectively 79% of the economics — hold ~14% of the vote.

The shareholder agreement also entrenches Vélez's control past the point where his stake might fall: as long as he beneficially owns 5% of the vote, he retains the right to nominate one director and veto debt issuance beyond consolidated net equity. Below 25% he still gets two seats; below 40%, three. This is hard-coded; it is not a renewable agreement.

Insider activity — sales, not buys

There has been no open-market insider buying disclosed in the period covered. Vélez's August 2025 33-million-share sale (~$462M at the day's $14 price) was the largest single insider transaction and triggered the 13G/A amendment showing Rua California's Class A stake stepping down. Junqueira added a ~$4M sale in March 2026. The April 2026 round of small "tax-withholding" dispositions on RSU vesting is mechanical and not signal-bearing on its own.

Read. The dollar magnitude is small relative to Vélez's remaining ~$11B stake — diversification, not exit. But the directional asymmetry (founders sell when they monetize equity; they never add) is normal for a still-young IPO and gives the bear case its only real ownership-trend talking point.

Dilution

Stock-based compensation under the 2020 Omnibus Plan and the founder Contingent Share Award programme remains the principal source of new Class A share issuance. The CSA was structured to pay Vélez additional shares only if performance hurdles are met; the rest of the team receives annual RSU grants. Total share count growth has been modest relative to the equity comp percentages typical at LatAm fintech peers — the bigger dilutive risk is the convertibility of Vélez's 901M Class B shares, which on conversion would become Class A but would not increase the total share count; it would, however, eventually collapse the voting structure on transfer.

Related-party behaviour

The 20-F discloses a registration-rights agreement with early holders, indemnification agreements for officers and directors that are broader than Delaware norms, and a related-party transactions policy updated in March 2024 requiring board or Audit & Risk Committee pre-approval. No material related-party contracts (consulting fees to founders, controlled-vendor agreements, family employment) are disclosed for FY2025. Sequoia (Doug Leone is still on the board) reduced its 13G position by 57% in May 2025 — that is a fund-cycle exit by an early backer, not a self-dealing event, but the optics of an insider VC unwinding while public holders sit through a 28% YTD decline deserve flagging.

Capital allocation

The board has not authorised a buyback. FY2025 net income of $2.87B has been redeployed into LatAm credit growth and the US/Mexico bank build-out. Capital allocation discipline rests entirely on Vélez and do Lago since the board has no meaningful counterweight on routine decisions.

Skin-in-the-game score

8/10. Vélez and Junqueira together hold ~$12.5B of NU equity at the current $12.18 price — among the highest absolute founder stakes of any newly-listed bank globally. The score is penalised two notches because (a) the dual-class structure means economic skin is decoupled from voting skin in a way that disenfranchises Class A holders, and (b) the only insider trading flow disclosed is selling, not buying, even after a sharp YTD decline.

4. Board Quality

Nine directors. Eight are independent under NYSE rules. Average age 56. Six committee chairs/members include a Brazilian CPA running Audit & Risk and a US bank CEO running Compensation & People — high-quality choices for a regulated LatAm bank.

Scores: 3 = deep expertise, 0 = none.

What the matrix shows. Two genuine banking-regulator brains (Calderón, Moreno) plus one ex–wealth-management COO (Sands) cover the audit and regulatory burden. Two deep engineering operators (Pham, Marcus) plus a recent additon (Piacentini, ex-Amazon International) give product-scale credibility. The gap: only Reses has hands-on US bank-charter experience — relevant because the company is now staffing up to launch a US bank. Outside Vélez/Junqueira, nobody on the board operates a LatAm-domiciled financial institution today, which matters for things like Brazilian central-bank examination handling and Mexico licensing posture.

Can this board challenge management? Probably yes on most operational decisions — Reses, Sands, and Calderón have all chaired or audited regulated entities and would not rubber-stamp risky underwriting. But on strategic decisions (M&A, where to deploy $2.9B/yr of profit, when to convert Class B shares), the shareholder agreement gives Vélez veto rights below even a 10% threshold. The board can advise; it cannot override.

5. The Verdict

Governance grade: B−.

The case for trust. A genuinely founder-led bank, 13 years into the build, with the founder's personal $11B stake still on the table. A senior management bench of ex–McKinsey/Itaú/HSBC/Google operators that grew through the IPO, the 2022 LatAm risk re-rating, and a recent profitable normalisation. A board that — on paper — has more independence and more relevant expertise than the controlled-company status would require. Compensation is heavily equity-weighted and well under 1% of revenue. There are no disclosed related-party shenanigans, no compliance restatements, no whistleblower events, no SEC or CVM enforcement matters.

The case against trust. Outside shareholders own ~80% of the economics and ~14% of the vote. Vélez's veto rights on debt, M&A, and major actions persist down to a 5% stake. Foreign Private Issuer status means no individual NEO comp, no Say-on-Pay, no Form 4. Cayman incorporation plus broad indemnification provisions make derivative litigation nearly impossible. Insider activity since IPO is sell-only, and the August 2025 $462M Vélez sale set the public-market tone for a stock that has fallen 28% YTD. Hiring Brazil's just-departed central-bank president as Executive Vice-Chairman is operationally smart but a regulatory-optics liability.

Upgrade trigger. Voluntary collapse of the dual-class structure on a defined sunset (e.g., the seventh anniversary of the IPO is December 2028), or the appointment of a fully independent Lead Director with explicit board-level authority to call meetings without the chairman, would re-grade the file to B+ / A−.

Downgrade trigger. A material related-party transaction (Rua California financing arrangement, founder-affiliated vendor contract), accelerated Vélez selling beyond diversification scale (more than 5% of remaining stake in any 12-month period), or any regulatory action involving Campos Neto's prior central-bank decisions that benefited Nu would drop the grade to C+ / C.

The Narrative Arc

Nu's story has bent without breaking. From IPO-day skepticism in late 2021 — when underwriters cut the price range from $10–$11 to $8–$9 days before pricing — the company has compounded customers from 54M to 131M, swung from a $364M 2022 loss to $2.9B of 2025 net income, and never changed founders. The founder-CEO who promised "the world's largest digital banking platform outside Asia" in 2021 is now selling a "rebuild banking around AI" thesis and a U.S. bank charter. Management has earned the right to be heard, but the next chapter — global expansion and AI transformation — is the first where the bear case lives in the future, not the past.

The chapter that defines the current franchise began in 2023 — the year the loss-making challenger became a structurally profitable retail bank. Everything before that is the asset; everything after is the test of whether the team can deploy it.

What Management Emphasized — and Then Stopped Emphasizing

Four discrete eras show up clearly in how management talks about the business. The phrase that anchors each year-end letter has rotated almost on schedule, and what was loud in 2022 ("path to profitability") is now almost silent, replaced by "AI transformation" and "global digital banking platform."

Management emphasis on each topic, scored 0 (silent) to 5 (anchor theme).

Three patterns matter:

Loud, then silent. "Path to profitability" disappeared the moment profitability arrived. So did "crypto" — promoted heavily in 2022 when NuCrypto launched, now a back-page mention. Insurance and NuInvest get one paragraph in the 2025 letter where they got pages in 2021. Management isn't hiding these — they simply stopped being the story.

The quiet swap. Through 2023, the headline number was customers. From 2024 forward, the headline is ARPAC per customer ($11.2 → $15.9 in seven quarters). This is the right pivot — saturated markets need monetization, not more signups — but it's worth noticing that it coincided with Brazilian net adds slowing from 25% to 14% YoY. The new metric flatters the new reality.

The two big additions. "AI transformation" went from a buzzword in 2024 to the anchor of the 2026 narrative, with nuFormer disclosed at parameter-count scale comparable to Meta's LLaMA. "US expansion" went from non-existent to the closing-remarks centerpiece of every 2025–26 call. These are real strategy shifts, not just framing.

Risk Evolution

The 20-F risk factor section is remarkably stable on the macro and regulatory front — Brazilian rate volatility, government intervention, and cybersecurity have been top-five risks every year. What has changed is what management worried about most: the existential "we may never be profitable" language has disappeared, and three new risks have surfaced.

Risk-factor emphasis in 20-F filings, scored 0 (absent) to 5 (top-tier).

What got smaller. The 2021 prospectus carried explicit "we have incurred losses since our inception, and we may not achieve profitability" language in its risk-factor summary. By 2024 that bullet point is gone from the summary section. The risk was retired by execution.

What got bigger. In 2024 and 2025, management added a brand-name risk — Brazilian regulators were consulting on whether non-banks could keep using the word "bank" in their name. That's a regulatory threat to "Nubank" itself, and it didn't exist in earlier filings. The 2025 20-F adds two new top-tier disclosures: AI operational/reputational risk (the first time it appears at all) and explicit Brazilian election risk tied to October 2026.

What stayed loud. Brazilian rate and FX volatility — the single most-cited macro risk — has not faded. Founder voting control has been a Class B risk every year since IPO, and Vélez's stake has shifted only marginally (88.6% → 88.3% of Class B).

How They Handled Bad News

Three episodes are worth comparing because the style of the disclosure tells you something about credibility.

Guidance Track Record

Nu does not give point-form earnings guidance, which makes "promises" harder to grade. But management has put specific, valuation-relevant claims into the public record over the years. Here is the scorecard on the ones that mattered.

Credibility score: 8 / 10. Every binary promise that's already had time to mature — profitability, ROE, Mexico break-even, efficiency-ratio compression — has been delivered, often ahead of internal plans (Q1 2026 IFRS-profit milestone "arrived ahead of our own internal plan," per Vélez). The two big TBDs — the "less than 100bps efficiency drag" U.S. cap, and the asset-quality narrative under AI-accelerated underwriting — are both future tests, not past misses. The CSA termination is the single strongest signal of CEO alignment in the file. The point off is for the Q1 2026 EPS miss, the introduction of a non-IFRS managerial framework as the primary headline, and the fact that the hardest promises (US, AI economics) haven't been tested yet.

What the Story Is Now

The 2026 story is in three parts: a profitable Latin American core, an asymmetric U.S. option, and an AI-transformation bet on whether banking gets re-platformed.

Believe. The Latin American franchise is the rare combination of high growth, structural cost advantage, and proven credit underwriting through a 110-bp NPL cycle. The team that built it is the team that still runs it. Promises about profitability, ROE, efficiency, and Mexico break-even have all been delivered, on or ahead of schedule.

Be skeptical of. The "AI-first bank" framing and the U.S. expansion claim of bounded downside. Both are forward-looking, both are unfalsifiable in the short term, and both arrived in the disclosure just as the "path to profitability" story retired. That timing — replacing one narrative pillar with two new ones precisely when the old one stopped working as a forward driver — is the pattern to watch.

The story is simpler than it was in 2022 (when the company was a customer-count chart with no earnings). It is more stretched than it was in 2024 (when Brazil execution was the whole thesis). Where it goes from here depends less on whether the LatAm flywheel keeps spinning — it will — and more on whether the U.S. and AI bets compound the franchise or merely consume capital while the core matures.

Financials in One Page

Nu Holdings is no longer a digital-bank story you have to fund with hope. In FY2025 the company generated $11.2B of revenue (+26.8% YoY), $3.87B of operating income (34.6% margin), and $2.87B of net income (25.6% margin). Operating cash flow hit $3.5B and free cash flow $3.49B — capex is essentially zero on the cash flow statement, so OCF and FCF are nearly identical, converting at roughly 1.2× of net income. Shareholders' equity has compounded to $11.3B and book value per share to $2.30, with ROE at 30.3%. The balance sheet is asset-heavy ($74.9B), but that is loan book and deposit-funded liquidity, not corporate leverage; long-term debt is just $4.4B against $24.5B of cash. The unsolved problem in FY2026 is valuation: at roughly 28× trailing earnings and 7× book, Nu is priced like a software company while it underwrites Brazilian and Mexican consumer credit. The metric that matters most right now is net interest margin (NIM) and the related early-stage NPL ratio — Q1 2026 showed NIM compressing from 10.5% to 9.5% and 15–90-day NPLs ticking up to ~5%, the signal that arbitrates the credit-cycle question.

How to read these. Revenue for a digital bank = net interest income + fees, before credit-loss provisions. Operating margin compares pre-tax operating income against that gross revenue line. ROE measures profit against shareholders' equity — for banks, this is the headline quality metric. P/B is price relative to book equity per share; banks typically trade between 1x and 2x P/B unless ROE is exceptional.

---

Revenue, Margins, and Earnings Power

Nu's revenue line is net interest income (interest earned on credit cards and personal loans minus interest paid on deposits) plus non-interest income (interchange, fees, asset-management). Provisions for credit losses sit below this revenue line — they are a real cost of doing business, but the company reports revenue before taking them. Watch all three lines together.

Revenue trajectory: from $0.3B to $11.2B in seven years

Two things to register. First, the revenue mix has shifted decisively toward interest income: NII was 43% of revenue in 2018; it is 79% in 2025. This is what happens when a credit-card platform turns into a full-stack consumer bank — the asset side (loans, cards) compounds faster than the activity side (interchange, fees). Second, the revenue ramp is non-linear: the 5-year CAGR is 73%, but the 1-year FY2025 growth is 27%. That deceleration is normal as the base grows, but it is also exactly why valuation discipline matters now.

Margins inflected in 2023 and have kept expanding

FY2023 was the cleavage. Up to FY2022 Nu was investment-mode: spending on customer acquisition, building Mexico and Colombia, absorbing SBC, and provisioning aggressively for a maturing credit book. From FY2023 onward, operating margin has stair-stepped from 25.7% → 32.2% → 34.5%, and net margin from 17.2% → 22.7% → 25.6%. Two drivers: SG&A grew only 25% from 2023 to 2025 while revenue grew 87%, and the credit book scaled past its provision burden so each incremental dollar of NII drops disproportionately to operating income.

Recent quarterly trajectory — operating margin holds, NIM compresses

Eight straight quarters of sequential revenue growth and operating margin sustained near 38% by Q4 2025. The flag is what management disclosed for Q1 2026: NIM stepped down from 10.5% to 9.5% as the loan mix shifted toward lower-yielding, lower-risk public payroll loans, and 15–90-day NPLs rose to ~5% on Q1 seasonality. The Q1 2026 print missed EPS by a penny ($0.18 vs $0.19 consensus) on higher provisions even as revenue beat at $4.96B. Margin durability through FY2026 is now the live question.

---

Cash Flow and Earnings Quality

For a bank, the most useful cash-quality check is operating cash flow vs net income. Capex is trivially small for a digital-only model (no branches, no ATMs, no fleet), so OCF and FCF are essentially the same number. Free cash flow is the cash the business generates after operating needs and the (tiny) capex bill.

OCF and FCF track and exceed net income

Three observations. First, OCF/NI ratio is 1.22x in FY2025, in line with FY2024 (1.22x) and FY2023 (1.23x) — earnings are converting to cash at a stable, above-100% rate, which is what you want to see in a maturing financial company. Second, capex is $7M on $3.5B of OCF (0.2% of revenue): there is essentially no capex drag, which is why this business throws off so much cash relative to a branch-based bank. Third, FY2021 OCF was deeply negative ($-2.9B) because Nu was funding hyper-growth in the loan book through working capital — that pattern has reversed completely now that deposit funding outpaces loan growth.

FCF margin is at 50% — a software-like print on a bank-like business

FCF margin of 50% in FY2025 is the kind of number you see at mature software companies, not banks. The 2020 spike and 2021 trough reflect working-capital swings around the credit-card receivables cycle, not real economics. From FY2022 onward the trend is steady upward — 40% → 34% → 43% → 50% — driven by SG&A leverage and a more disciplined provisioning cadence.

Earnings-quality watchouts

SBC at 2.4% of revenue is reasonable for a tech-led financial; the line to watch is provisions, which scaled to $4.2B in FY2025 (38% of revenue before provisions). That is the number most exposed to a cyclical credit downturn.

---

Balance Sheet and Financial Resilience

For a bank, balance-sheet analysis is different. Most "liabilities" are customer deposits — the cheapest funding in finance, not corporate debt. The risk is not interest coverage; it is asset quality, capital adequacy, and funding mix.

Liquidity, deposit base, and equity have all scaled together

Net cash position is consistently large: $24.5B of cash against just $4.4B of long-term debt at year-end FY2025. Equity has compounded 23x since FY2018 — partly because of the December 2021 IPO ($2.6B equity raise), and partly because retained earnings have flipped from accumulated deficit to $6.4B of retained earnings. Long-term debt is rising (from $1.1B in 2023 to $4.4B in 2025), but this is wholesale funding of the loan book at the holding-company level, not distress; assets-to-equity dropped from 23x (FY2020) to 6.6x (FY2025), the safest leverage profile in Nu's history.

Bank-specific resilience: capital, asset quality, and funding

Three things matter here. Loan-to-deposit at 58% means Nu is still funded with substantially more deposits than it has lent — a fortress position that lets the credit book grow another 30–40% before funding gets tight. Cash at 33% of assets is far higher than any incumbent Brazilian bank — Nu deliberately runs hot on liquidity. Goodwill / assets at 0.5% confirms growth is organic; there is no large M&A balance-sheet risk to mark down.

Asset quality is the real watch item

The 15–90-day NPL ratio rose to ~5% in Q1 2026 (up 89 bps from year-end on seasonality), and management said NIM will stay closer to the 9.5% Q1 level than the 10.5% Q4 level near term. Coverage is over 150% of NPL formation, which is conservative, but the trajectory of provisions ($4.2B in FY2025, up 33% YoY) tells you Nu is still front-loading credit costs against a fast-growing book. A genuine Brazilian credit cycle would test this.

---

Returns, Reinvestment, and Capital Allocation

Three questions: is ROE high and rising? Is the company reinvesting at high incremental returns? Is per-share value compounding faster than the share count?

ROE has moved from -33% to 30% in five years

ROE at 30.3% in FY2025 is more than double what Itaú prints in a strong year, and 5x what Bradesco managed. ROA at 4.7% is also exceptional for a bank — a function of the asset-light digital model that doesn't carry branch real estate. Management has guided ROE to remain around 30%+ in Q4 2025; whether that holds through a credit downturn is the real test.

Capital allocation: organic reinvestment dominates, buybacks are minimal

Almost everything Nu earns is going back into the loan book — that is the reinvestment story. There is no dividend, no material buyback ($6.2M in FY2025), and no M&A ($1.5M). The shareholder yield is essentially zero by design. Whether that is the right call depends on whether Nu can keep earning 30%+ ROE on retained capital — at present, the math supports the choice, but a maturing customer base in Brazil will eventually push management toward returning cash.

Share count: dilution has slowed sharply

The 2021–2022 jump was the IPO. Since then, dilution has been under 1.5% per year, dominated by employee stock vesting. SBC as % of revenue fell from 33% (2022) to 3.9% (2025). For a tech-flavored bank this is well-controlled.

Per-share value compounding

Book value per share has doubled in three years, and tangible book is barely below reported book — goodwill is small and intangibles are real (software, customer relationships). With ROE near 30%, book should keep compounding at high-teens to mid-20s annually if growth holds.

---

Segment and Unit Economics

Nu reports as a single operating segment: digital banking and financial services. Geographic disclosure is similarly thin (Brazil dominant; Mexico and Colombia small but growing). The most useful unit economics come from management commentary, not segmented financials.

The economics that matter: monthly ARPAC of ~$16 in Q1 2026 vs ~$11 a year earlier — a 45% per-customer revenue lift while the customer base grew 24%. The cross-sell engine is alive. Mexico reaching break-even is the single most important strategic milestone of FY2026 — it proves the Brazilian playbook ports. Brazil at ~7% of the addressable profit pool is management's framing for why this is still an early-innings story even at 115M customers.

---

Valuation and Market Expectations

The valuation debate is binary. Bulls say Nu deserves a premium because of 30%+ ROE, 50% FCF margin, and 25%+ revenue growth. Bears say no bank, even an excellent one, deserves 28x earnings and 7x book in a country with a structurally high cost of capital.

Current multiples vs Nu's own history

Two history lessons. First, Nu was unprofitable in FY2021–FY2022 so trailing P/E was meaningless; the real anchor is FY2023 onward, where P/E has compressed from 39x to 28.6x as earnings have grown faster than price. Second, P/B has expanded from 3.9x (FY2022) to 7.2x (FY2025) even as book value has more than doubled — the market is awarding a richer multiple to richer book. This works only if ROE stays above 25%.

What the market is pricing

The sell-side is leaning bullish. Median price target is $19.50 versus a current price near $13, implying ~50% upside. The forward P/E at ~17x on FY2026 EPS doesn't look extreme on consensus growth (+52% EPS), but the comp set of mature banks trades at ~11x. The premium is paid for digital scale and ROE persistence.

Bear / Base / Bull frame

At ~$13 today, you are roughly halfway between the bear and base cases — a market that respects the growth but no longer pays for blue-sky optionality. The risk/reward is asymmetric to the upside if credit holds and Mexico scales, and asymmetric to the downside if Brazil's consumer cycle turns.

---

Peer Financial Comparison

A bank's peer comp lives or dies on three metrics: ROE, revenue growth, and valuation. Currency and balance-sheet structure differ across the comp set, so the comparison is conceptual, not line-by-line.

BRL-reporting peers (ITUB, BBD, BSBR, STNE, PAGS) converted at ~5.0 BRL/USD for revenue/net income; market caps already in USD. Peer ratios are approximate FY2025 figures from each company's snapshot.

The grid tells a clean story. Among Brazilian banks, Nu is the only name combining 25%+ revenue growth and 30%+ ROE — Itaú leads on absolute net income ($8.8B), but its ROE is 20.5% and growth ~10%. Bradesco and Santander Brasil are in single-digit ROE territory; their cheap multiples (1.2x and 0.4x P/B) are deserved. Among fintech-style peers, MercadoLibre has higher growth (35%) but lower net margin and similar ROE, and trades at 41x P/E and 12x book — making Nu look reasonable. Among Brazilian fintech peers, StoneCo and PagSeguro trade at low multiples because their economics are weaker: PAGS at 15% ROE and 13% growth, STNE at 15% ROE and shrinking revenue.

The premium Nu commands over incumbent banks is earned by growth and ROE and not yet earned over MercadoLibre, which is the closest comparable on profile.

---

What to Watch in the Financials

The financials confirm that Nu has built a durable, high-quality earnings engine: 30%+ ROE, 50% FCF margin, dominant deposit franchise, and capital allocation that compounds book value at 20%+ per year. They contradict the simplistic "expensive bank stock" framing — at 28x trailing P/E the multiple is high, but at 17x forward it is defensible given the growth and profitability profile.

What they leave open is the credit cycle. Brazil has not had a serious consumer-credit downturn since Nu became dominant; the NIM compression and NPL uptick in Q1 2026 are the first whispers of pressure. If they turn into shouts, provisions can absorb $1–2B of operating income in a quarter and the multiple compresses to a true bank P/E in months.

The first financial metric to watch is the 15–90-day NPL ratio in Q2 2026: if it stays at or below 5% with coverage above 150%, the bull case keeps compounding; if it pushes through 6% with falling coverage, this stock re-rates fast.

Web Research

The Bottom Line from the Web

The filings show a company compounding revenue at 57% and customers past 135 million. The web shows that same company de-rated more than 35% from its $18.98 high on a single thesis-testing event: Q1 2026 credit-loss provisions jumped 33% QoQ, the early-stage NPL ratio jumped 89 bps to 5.0%, and risk-adjusted NIM fell 100 bps to 9.5% — first real-time evidence on whether NuFormer-driven underwriting holds up under stress. Wall Street is unmoved, with 18 Buy / 2 Hold / 1 Sell and a median target of $19.50 versus a market price near $12.

What Matters Most

Recent News Timeline

Analyst Snapshot

WSJ shows 18 Buy, 2 Hold, 1 Sell out of 21 analysts covering NU as of mid-May 2026. The dispersion is wide: low target $13.90 (essentially flat to current) versus high $22 (+80% upside). UBS upgraded to Buy on 2026-03-19; Zacks downgraded from Strong Buy to Hold on 2026-03-24 — the same window in which the stock began its second-leg decline.

Q1 2026 Snapshot — The Quarter That Set the Cycle Test

What the Specialists Asked

Governance and People Signals

Industry Context

The web evidence reinforces three industry shifts that materially shape Nu's near-term thesis beyond what filings disclose.

Brazil regulatory tightening on consumer credit. The November 2025 FGTS rule change drove a 50-60% decline in FGTS loan originations sector-wide (247wallst.com 2026-05-14). The existing 100% revolving-credit cap on credit cards continues to compress NIM. In December 2025, Brazil's Finance Minister Haddad proposed expanding BCB's mandate to supervise investment funds (Reuters 2025-12-17), part of a broader Brazilian credit-tightening cycle.

Brazil fintech corporate income tax step-up. Fintech corporate income tax rises progressively from ~40% to 45% beginning 2026. The Q1 2026 8.68% ETR is a transient interest-on-equity benefit; the structural rate normalizes higher across FY26-FY27.

Mexico is the validated second engine. Nu Mexico hit break-even in Q1 2026 at 15M customers (~14% of population vs. 60% in Brazil), now third-largest financial institution in Mexico. CNBV granted full banking-license approval; Nu continues operating as a SOFIPO during transition (stocktitan.net). Specialist research could not surface ARPAC comparisons against Plata, Ualá, or Revolut, so the head-to-head with other Mexico neobanks remains unresolved.

US national bank charter (conditional) is incremental and capital-light by design. OCC conditional approval on 2026-01-27 gives Nu, N.A. the framework for deposits, cards, lending, and digital asset custody. Management has capped efficiency-ratio drag at under 100bps per year through 2027 (chartmill.com 2026-05-15). Brian Brooks (former Acting OCC Comptroller) and Kelley Morrell (ex-Blackstone/Treasury) joining the US board signals regulatory seriousness without committing to a US retail strategy.

LatAm fintech as an asset class is de-rating. Tech-claude flagged the 28% YTD NU drawdown as ambiguous between NU-specific and broader EM/LatAm de-rating. Reuters coverage shows Brazil-specific political risk overlay (Senator Flavio Bolsonaro's presidential run, Haddad's BCB mandate expansion proposal), so part of the move is sector-wide rather than fundamental.

Web Watch in One Page

After Q1 2026, the Nu Holdings thesis stopped being theoretical and became testable. Through-cycle ROE on a $32.7B, 92%-unsecured book is now genuinely contested, the deposit-funding moat is about to be stress-tested by a Selic cut cycle, and the load-bearing optionality (Mexico, the OCC charter) is still priced at zero. The five monitors below sit directly on those open questions. They are designed to fire when the real-world data that arbitrates the bull/bear debate actually moves — Brazilian policy and election news that resets the regulatory backdrop, Q2/Q3 credit-quality prints that arbitrate cycle versus seasonality, competitive deposit migration through Open Finance rails, US/Mexico operational milestones, and insider and disclosure activity at the founder-controlled FPI. None of them ask for "latest news" — each is anchored to a specific variable the report has named as load-bearing or thesis-breaking.

Active Monitors

Rank	Watch item	Cadence	Why it matters	What would be detected
1	Brazilian regulatory, Selic policy, and Oct 2026 election	1d	Threads through three thesis variables at once — deposit cost via Selic path, unsecured-credit spreads via consumer-credit caps, and the effective tax rate via the JCP regime and the 40%→45% corporate-rate step-up	BCB Copom decisions and forward guidance; election polling and credible populist consumer-credit proposals (interchange, personal-loan, interest-on-equity caps); progress on the BCB mandate-expansion bill; further FGTS or revolving-credit rule changes
2	Q2 / Q3 2026 credit-quality print and Brazilian sector NPL data	1d	The single biggest update to whether Q1 2026's NIM compression (10.5%→9.5%) and NPL 15-90 spike (4.1%→5.0%) was seasonal mix or the leading edge of an unsecured-credit cycle the bears have called for two years	Nu Q2 (est. Aug 13-14) and Q3 (est. mid-Nov) 6-K releases with NPL 15-90, NPL 90+, Stage 3 coverage, risk-adjusted NIM, cohort ROA on NuFormer vintages; BCB SCR monthly household credit statistics; IBGE unemployment prints; sell-side FY26 EPS revisions
3	Competitive deposit migration via Open Finance	1w	Driver #1 of the long-term thesis. Deposit cost has held in an 88-91%-of-CDI band on a 2.2x scaling of the deposit base; any drift toward 92%+ as Mercado Pago Brazil, Inter, C6, and PicPay use Open Finance rails to lure balances breaks the funding moat	Mercado Pago Brazil deposit-base disclosures crossing $10-20B; competitor deposit-pricing or primary-relationship campaigns; BCB Open Finance adoption statistics; any Nu disclosure of deposit cost above 92% of CDI on the supplemental release
4	Nubank N.A. (US) and Mexico operational milestones	1w	The optionality the market currently prices at zero. A visible US product launch within the under-100-bps efficiency-drag cap validates the international compounding leg; a quiet capital injection without launch signals the US is a capital sink. Mexico has to clear ARPAC $15 with positive segment NI for the bull case to hold	OCC supervisory actions; Nubank N.A. product launch announcements; executive moves involving Cristina Junqueira, Roberto Campos Neto, Brian Brooks, Kelley Morrell; Nu Mexico ARPAC, customer-count and segment disclosures; new Mexican competition from BBVA Mexico, Plata, Ualá, Revolut, Mercado Pago Mexico
5	Insider activity and disclosure-machinery changes	2w	Founder-controlled FPI, ISAE 3000 limited assurance on the new Managerial P&L, $243M of FY25 capitalized intangibles, no insider open-market buy since IPO, Sequoia 13G cut 57% in May 2025. The machinery for catching aggressive judgment calls is running through fewer hands as the narrative pivots from "profitability delivered" to "AI + US charter"	SEC 13G/A, 13D, Form 144, Form 4 filings on Vélez, Junqueira, Doug Leone/Sequoia; FY2025 20-F filing with FY2024/23 Managerial P&L re-presentation; auditor assurance language; new non-IFRS add-backs; related-party transactions involving Roberto Campos Neto

Why These Five

The report names three load-bearing thesis variables — through-cycle ROE on the unsecured book, deposit cost as a percentage of CDI through a Selic normalization, and whether Mexico/US optionality earns its capital — plus two governance failure modes that move independently. Monitor 2 watches the credit-cycle variable in real time, Monitor 3 watches the funding moat as Selic begins to compress, Monitor 4 watches the optionality, and Monitor 5 watches the governance machinery. Monitor 1 sits upstream of three of those four — Brazilian regulatory and political outcomes drive Selic, the consumer-credit caps that hit the unsecured book, and the JCP and corporate-tax step-up that arbitrate FY26 EPS. The set is built to fire when a thesis variable actually changes, not when the next earnings date arrives.

Where We Disagree With the Market

The sharpest disagreement: consensus FY2026 EPS of $0.89 is mechanically too high because it extrapolates a one-quarter 8.7% effective tax rate that cannot persist into a Brazilian fintech regime where the corporate rate is stepping from ~40% to 45%. Sell-side carries a $19.50 median target on that EPS at ~17× forward; correct the denominator for a structural 22–27% ETR and the same multiple maps to roughly $0.70–0.75 of EPS — consistent with a $13–15 scenario, close to where the stock trades.

The market's 35% drawdown from $18.83 has been blamed on a credit-cycle scare, but the second disagreement is that the load-bearing 10-year variable is deposit cost as a percentage of CDI through a Selic normalization cycle, and consensus models assume the 88%-of-CDI funding edge holds as a flat dollar number rather than halving when Selic moves from 15% toward 9–11%.

The third disagreement is institutional rather than analytical: short interest at 2.9% of float says the de-rate is forced LatAm/EM redemption flow, not directional bears, which makes the marginal seller a price-insensitive holder rather than an informed one. Each resolves on observable data — Q2 2026 prints the tax rate, the BCB Copom path prints the funding-moat compression mechanism, and 13F flow into Q3 prints whether the marginal seller stays redemption-driven.

Variant Perception Scorecard

Time to resolution: 3–6 months (ETR), 12–24 months (funding).

The variant is strong but not maximal. Consensus is unusually clear — an 18/2/1 buy ratio with a $19.50 median target and a published FY26 EPS of $0.89 leaves little ambiguity about the market's implied model. The evidence base is solid: the Q1 2026 8.7% ETR is a disclosed fact, the Brazilian corporate tax step-up is regulator-confirmed, and the 88%-of-CDI funding base is in supplemental disclosure. What pulls the score below 80 is that the EPS-denominator critique is partially front-run by the stock at $12 already, and the funding-moat critique resolves over years rather than quarters — so the variant is correct but not maximally monetizable in a 90-day window.

---

Consensus Map

---

The Disagreement Ledger

Disagreement 1 — the EPS denominator. Consensus would say FY26 EPS at $0.89 is the right number because growth, operating leverage, and a 50%+ ROE on retained earnings carry the line. Our evidence disagrees because the 8.7% Q1 ETR is a one-quarter interest-on-equity event that cannot annualize — the Brazilian fintech statutory rate is rising to 45%, and management has signalled the structural rate "normalizes higher." If we are right, the market would have to concede that the same $19.50 sell-side target carries a 23-26x forward multiple, not 17x, and the "obvious upside to consensus" framing collapses to a fair-value-around-here framing. The cleanest disconfirming signal is Q2 2026 reported ETR — a print at or above 20% forces the cut; a sustained print below 15% would force the bear to explain how the structural tax rate avoided the regime that hit every Brazilian fintech peer.

Disagreement 2 — the funding moat is regime-dependent. Consensus would say the deposit franchise is the structural moat that makes the multiple defensible — 88% of CDI, sticky, primary-relationship, holds through Open Finance. Our evidence disagrees because the moat is a percentage, not a dollar — at Selic 9-11% the absolute funding gap halves even if the percentage holds, and the long-term thesis tab is explicit that this is the single load-bearing test for the 10-year case. If we are right, sell-side targets that implicitly assume 25%+ through-cycle ROE need to be reset to 18-22%, which compresses the multiple to 4-5x P/TBV — implying a $9-11 stock, not $19.50. The cleanest disconfirming signal is the quarterly deposit-cost-percentage-of-CDI disclosure once Selic cuts begin; if it stays in the 87-91% band on a $50B+ deposit base, the moat survived regime change and the bear loses this leg.

Disagreement 3 — the marginal seller is technical, not informed. Consensus would say a 35% drawdown from peak is the market doing its job — the bears have arrived. Our evidence disagrees: short interest at 2.9% of float and 3.41 days-to-cover is not consistent with directional bears writing the credit-cycle thesis. The de-rate looks like LatAm/EM redemption flow plus Brazilian political risk repricing, layered on a fundamental Q1 disappointment that gave the rotation an excuse. If we are right, a Q2 print that even partially refutes the credit-cycle read produces an asymmetric rally because there is no informed short cover waiting to absorb the move. The cleanest disconfirming signal is short interest spiking above 5% of float before Q2 — that would tell us informed bears have actually positioned and the technical-seller interpretation is wrong.

Disagreement 4 — the disclosure framework erosion is non-trivial. Consensus is taking the Managerial P&L at face value because net income reconciles. Our evidence is that the framework is on ISAE 3000 limited assurance only, FY24 and FY23 have not been re-presented, and $243M of FY25 development spend was capitalized rather than expensed under IAS 38. None of this is fraud, and Forensics keeps the file in the "Watch" band rather than "Elevated" — but the comparability and audit machinery is materially weaker than at a US domestic, and a PM paying 7x tangible book is entitled to a 100-200 bps multiple haircut for the asymmetry. The cleanest disconfirming signal is the FY25 20-F: a clean three-year re-presentation under the Managerial framework with no new non-IFRS add-backs would remove this concern.

---

Evidence That Changes the Odds

---

How This Gets Resolved

---

What Would Make Us Wrong

The strongest disconfirming evidence is the implied FY25 ROA of 4.58% versus Itaú's 1.47% — the single cleanest moat fact in the file. That gap is not a peak-Selic artifact; even if Selic normalization compresses it by 30-40%, Nu still earns ~3.0% on assets versus 1.0-1.5% at incumbents on roughly half the leverage. If our EPS-denominator critique is right but the ROA gap closes by less than 100 bps under Selic normalization, then the through-cycle ROE clears 22-25% rather than 18-22%, and the bull's $22 target on $1.20 of 2027 EPS becomes the right anchor — not consensus's $19.50 and not the bear's $8. We are betting that consensus is paying for permanence on a partly cyclical edge; that bet loses if the edge is more structural than cyclical.

The second place we are most likely wrong is the short-interest interpretation. Short interest data is reported with a two-week lag, and the cohort could already be building. If a fresh print shows short interest above 5% of float, the "technical seller" interpretation breaks — the de-rate becomes thesis-driven and the asymmetric-rally setup we are betting on disappears. We would also be wrong if the LatAm/EM derate stays intact through Q2 regardless of fundamentals; in that case even a clean Q2 print gets sold and the variant pays out on a longer horizon than we think.

The third risk is that the interest-on-equity (juros sobre capital próprio) regime extends longer than the corporate-rate step-up suggests. Brazilian tax policy is politically negotiated. If the JCP benefit is preserved through the October 2026 election and into 2027, the 8.7% ETR could persist for 3-4 quarters rather than 1, and consensus EPS at $0.89 could prove more durable than the variant assumes. The bear-case-on-the-variant is that we are right on the structural rate but wrong on the path, and the EPS revision we expect lands in 2027 rather than 2026 — which materially weakens the 90-day monetization case.

The fourth risk is the Mexico optionality call. We have flagged it as priced at zero, which is favourable for our variant. But Mexico break-even is one quarter, and the competitive density there (BBVA, Banorte, Plata, Ualá, Revolut, Mercado Pago) is materially higher than Brazil 2014-2018. If Mexico ARPAC stalls below $10 through 2027 even with continued customer growth, the "second engine" framing fails to compound, and the variant view of "optionality undervalued at zero" loses its kicker — the case has to win on Brazil alone.

The first thing to watch is the Q2 2026 effective tax rate, because it is the single line item that mechanically arbitrates whether consensus FY26 EPS of $0.89 survives the print or gets cut by 15-20% in a week.

Liquidity & Technical — Nu Holdings (NU)

Implementable at small size with deep ADV, but the tape is broken: NU sits 21% below its 200-day moving average, printed a fresh death cross on 15 April 2026, and closed at $12.18 — barely 4% above the 52-week low. The dominant feature is a six-month price decline of roughly 28% YTD on rising daily ranges, with RSI at 25 (oversold) and MACD histogram deepening into negative territory. The tape is bearish; execution can wait for a base rather than chase weakness.

1. Portfolio implementation verdict

2. Price snapshot

The stock sits in the bottom 4% of its 52-week range. Beta is not provided in the staged dataset; readers should treat this as a high-volatility EM financials name (30-day realized vol of 32%).

3. Price history with 50/200 SMA

Price is below the 200-day moving average ($12.18 vs SMA200 of $15.47 — a 21.2% discount). The most recent 50/200 cross was a death cross on 15 April 2026, undoing the 7 July 2025 golden cross in under nine months.

The full-history chart is a story of three regimes: a deep IPO-era drawdown into mid-2022 (price below $4), a multi-year recovery base that broke out in 2024 to all-time highs near $18.83 in early 2026, and an aggressive eight-week unwind that has erased the entire 2026 melt-up. This is a downtrend regime.

4. Relative performance (rebased)

The staged data file contains a NU index rebased to 100 at the start of the three-year window but does not include benchmark or sector ETF series (broad_market and peers_in_basket are empty in relative_performance.json). The chart below shows the absolute three-year return profile rebased to 100; readers wanting a true RS line versus SPY or LatAm financials should treat this as a placeholder.

NU peaked at 313 (index) in late January 2026 — a triple over three years. From that peak it has shed 34% in 15 weeks. The shape is a classic momentum unwind, not a slow distribution: the rate of decline has accelerated since mid-March.

5. Momentum — RSI(14) and MACD histogram

RSI closed at 25 on 13 May 2026 — the most oversold reading in the entire 18-month window, deeper than the December 2024 trough (24.2). MACD histogram has flipped negative again after a short rally in early April; both signal and line are below zero with the gap widening. Near-term momentum is bearish but stretched — the kind of setup that produces a tactical bounce without changing the trend. Watch for a positive divergence (price making new lows while RSI prints higher lows) before adding.

6. Volume, volatility, and sponsorship

Volume swelled to 130M shares on 15 August 2025 (the post-Q2 print) and again touched 79M on 2 March 2026 — both sessions saw price decline, suggesting distribution rather than accumulation. The 50-day average compressed through Q4 2025 (low-30M range as the stock melted up) and has snapped back into the high-40M as the decline gathered force; rising volume into a downtrend is bearish confirmation.

Top 3 volume-spike days (all history)

Realized volatility, five-year context

Realized vol at 32% sits just inside the "calm" band (sub-p20 = 33.6%). The April–May drawdown moved on modest, not panic-grade volatility — which means the selling has been orderly and institutional, not retail capitulation. Calm-regime declines tend to require either fundamental thesis change or charted base-building before they reverse; option-implied vol cheap-to-realized here would favour buying tail protection over chasing rebounds.

7. Institutional liquidity panel

NU is a $59.8B mega-cap with $501M ADV (20d) and 255% annualized share turnover — the book is deep. The constraint is not getting in or out; it is how large a single fund position can be without bleeding into the tape.

Fund-capacity scenarios

A fund running at 20% of ADV can execute roughly $435M in five sessions — that is 5% of an $8.7B portfolio. The same fund running at a more conservative 10% participation clears $218M, capping a 5% position at $4.4B AUM. Concentrated mid-sized funds (under $5B) can establish a full position in a week; large multi-strategy or sector funds will need staged execution over several weeks.

Liquidation runway for issuer-level position sizes

Median daily range over the last 60 sessions is 1.28% — below the 2% elevated-friction threshold, so intraday price impact is normal. A 0.5% market-cap position (about $299M) clears in four sessions at 20% participation; the same position takes a week at the more conservative 10% rate. Largest issuer position that clears the five-day threshold is 0.5% of market cap at 20% participation — anything bigger requires multi-week staging. Net: liquidity is not the bottleneck for sub-$300M positions; it is a soft constraint above that.

8. Technical scorecard + stance

Stance: bearish on the 3–6 month horizon. Net score of −4 across six dimensions, with no offsetting positives. The setup is a freshly broken uptrend with deep momentum oversold but no evidence of a bottom — both classic conditions for a tactical bounce that fails at the 200-day. Invalidation levels: a daily close above $15.50 (the 200-day moving average and prior support shelf) would force the view to neutral and re-open the bullish case; a daily close below $11.90 (the 52-week low) confirms the next leg lower and opens $10 / $9 as the realistic re-test zone. Liquidity is not the constraint — for any fund considering this name, the correct action is watchlist only until either level resolves, then build slowly if and only if the reclaim is on expanding volume. Chasing the current oversold print is a mistake.