Analytical Thesis
Nu Holdings is executing one of the most capital-efficient bank-building exercises in financial history. By Q1 2026, the company reached 135.2 million global customers, surpassed $5 billion in quarterly managerial revenue for the first time, and reported $871 million in net income — nearly doubling year-over-year. The efficiency ratio hit 17.6% in Q1 2026 (down from 21.4% a year prior), placing Nu structurally ahead of every global banking peer. Monthly ARPAC reached $15.90 (+23% YoY), while the monthly cost to serve remained at ~$0.90 , well below the critical $1.00 threshold. With a $0.80-0.90/month cost to serve and 17%+ NIM, the company sits at the inflection point between high-growth digital challenger and mature compounder.
Nu is often framed as a steady-state bank. It is not. The more useful question is what a business with 150M+ customers, 4+ products per customer, and platform optionality across Latin America’s 650M-person underbanked population can become at maturity.
How Nu Makes Money
Nu Holdings is a vertically-integrated digital financial services platform operating across Brazil (101.6M customers), Mexico (10M), and Colombia (2.5M). Revenue flows through three mechanisms: (1) net interest income from credit card revolving balances, personal loans, and government bond investments; (2) service fees from card interchange, account management, and financial services; and (3) emerging fees from investment products, insurance, NuCel (MVNO), and NuTravel.
The business is fundamentally a bank operating with the cost structure of a technology company.
Nu cost per customer/mo
$0.80
lowest in financial history
Itau cost per customer/mo
~$12
15x Nu's cost base
Nu efficiency ratio
29.9%
vs. 50-65% for global peers
Monthly cost to serve per active customer. Nu at $0.80 is structurally disconnected from traditional banking -- as ARPU grows, nearly all incremental revenue is margin.
FY2025 Statement-Level Deep Dive
Business from the Ground Up
Nu Holdings (Nubank) is executing one of the most audacious and disciplined financial platform builds in history. What began as a no-fee credit card distributed entirely digitally in Brazil (2014) has become the largest digital bank in the world by customer count — 131 million customers across Brazil, Mexico, and Colombia as of year-end 2025 — and, importantly, one of the most profitable. The central insight behind Nubank is structural: Latin American incumbent banks, protected by regulatory oligopolies and distribution moats for decades, had never been forced to compete on cost, experience, or transparency. Their efficiency ratios of 50–70% reflected a captured market, not operational excellence. Nu entered with a technology-first, zero-branch cost structure and built a financial services platform where the monthly cost to serve an active customer is $0.80 — versus incumbent cost-to-serve of $20–25. The 25–30× cost advantage is the enduring competitive moat. It cannot be replicated by incumbents without dismantling their existing branch network, workforce, and core banking infrastructure — a multi-decade, billion-dollar, culturally impossible transformation.
Credit Card (core originating product, largest revenue contributor) The credit card remains Nu’s customer acquisition vehicle and the largest asset on its balance sheet. Revenue is earned through: (1) interchange fees paid by merchants (~2% of transaction value in Brazil), (2) interest income on revolving balances, and (3) late fees. The credit card product was intentionally loss-leaderish in the early years — no annual fee, cashback, clean mobile UX — designed to maximize customer acquisition and data accumulation, with monetization to follow as Nu demonstrated creditworthiness data superiority over incumbents. Brazil’s consumer credit market is structurally expensive for borrowers (benchmark lending rates above 40% APR for revolving credit) — Nu disrupts not by charging close to zero, but by charging significantly less than incumbents while maintaining risk-adjusted NIM of 10.5% in Q4 2025, indicative of a disciplined underwriting process as the portfolio matures.
Personal Loans, Payroll Loans, Buy-Now-Pay-Later, and Secured Credit After the credit card, the natural product expansion is unsecured personal lending (already launched and scaling), payroll-deducted loans (consignado — lower risk, lower rate, high volume, launched in Brazil in 2025), and secured credit products. Each new lending product expands the addressable market per customer — cross-sell is the core monetization strategy. The NIM dynamics of payroll loans are differentiated: because repayment is deducted automatically from the employer payroll, default risk is materially lower, allowing Nu to price at lower rates while maintaining adequate risk-adjusted returns. NIM company-wide was approximately 17.3% in Q3 2025, with risk-adjusted NIM of 10.5% in Q4 2025 (and would have been broadly stable sequentially ex a one-time Prosofipo contribution in Mexico).
Savings Accounts and Deposits (funding franchise) Nu launched instant-return savings accounts early in its lifecycle — a critical strategic move because it converted the product from “credit card provider” to “primary banking relationship.” Total deposits reached $41.9 billion in Q4 2025, up 29% YoY FX-neutral. In Brazil, Nu pays approximately CDI (the Brazilian benchmark rate) on deposits — a competitive rate that made Nu the obvious alternative to incumbent banks who paid below-CDI rates on savings. The deposit base provides Nu with a low-cost, diversified, and growing funding source for its loan portfolio, directly improving risk-adjusted NIM as the deposit-to-loan ratio improves. Brazil contributed $33.2B, Mexico $6.3B, and Colombia $2.5B of total deposits in Q4 2025.
Investment Platform (NuInvest / Easynvest) Nu acquired Easynvest in 2020 and rebranded it NuInvest, offering ETFs, fixed income, and equity investments through the Nubank app. The investment platform is currently a secondary revenue stream but serves the critical strategic function of deepening the primary banking relationship: a customer who invests through Nubank is orders of magnitude more likely to stay, open additional products, and increase card spending than a customer who only holds the credit card. Monthly ARPAC (Average Revenue per Active Customer) reached $15 in Q4 2025, up 18% FX-neutral YoY — and this figure understates the monetization opportunity, since the fully-banked, multi-product Nu customer generates far above average ARPAC while cost-to-serve remains approximately constant.
Insurance (NuSeguros) and Other Services Life insurance, auto insurance, and homeowners insurance are distributed through the Nubank app, generating referral fees and commission revenue. Insurance penetration in Brazil and Latin America is materially below OECD averages — a long runway for growth as Nu’s increasingly affluent customer base seeks protection products that historically required a bank branch or insurance agent relationship. The insurance segment also benefits from Nu’s data advantage: behavioral and transactional data enables better risk assessment than traditional actuarial models.
Geographic Expansion: Mexico and Colombia Mexico (15M customers, ~15% of adult population) is at an earlier stage of the Brazil maturity curve — Nu is the leading new credit card issuer in Mexico, building a deposit base ($6.3B in Q4 2025), and recently began payroll-deducted lending. The Mexico unit economics are improving on both Y/Y and Q/Q basis ex-Prosofipo. Colombia (4M+ customers) is approaching an inflection point, with credit card approval rates nearly tripled following recent portfolio expansion. The international operations (Mexico + Colombia) are investments today — absorbing capital and generating early-stage losses — but following the Brazil template: acquire customers at low cost, prove creditworthiness data, cross-sell deposits and lending, expand NIM as the portfolio matures.
Income Statement
Nu’s FY2025 income statement represents the transition from high-growth digital challenger to structurally profitable financial institution — at a pace that has no parallel in global banking history. FY2025 revenues reached $16.3 billion, up 45% YoY FX-neutral — a growth rate on a $16 billion base that would be remarkable in software, let alone in a regulated financial institution exposed to Brazilian FX and credit cycles. Gross profit reached $6.6 billion in FY2025 versus $5.0 billion in FY2024. Net income was $2.9 billion in FY2025 (+45% YoY), with Q4 2025 delivering a record $895 million (+50% YoY), and ROE reaching 33% — a number that places Nu among the top-performing financial institutions globally on this metric. The efficiency ratio — non-interest expenses as a percentage of net revenue — declined to 19.9% in Q4 2025, down from 78% in Q4 2021. No incumbent bank in Latin America achieves below 40%; most operate between 50–65%. This 30–45 percentage point structural advantage in efficiency is the operating leverage story: as Nu’s revenue base grows, non-interest expenses (predominantly technology infrastructure and people) grow at a fraction of the rate.
Net Interest Income (NII) reached $2.837 billion in Q4 2025, a new all-time high, growing 13% QoQ on a seasonally strong quarter (Brazilian 13th salary effect). NIM of 17.3% (Q3 2025) with risk-adjusted NIM of 10.5% (Q4 2025) reflects a portfolio that, while exposed to Brazilian consumer credit cycles, is increasingly diversified across products (credit card, personal loan, payroll loan), geographies, and customer segments (mass market, affluent Ultravioleta tier, SME). The 15–90 NPL ratio improved for the fourth consecutive quarter in Q4 2025 to 4.1%, and 90+ NPLs declined 10bps to 6.6% — tracking within expected ranges and consistent with the seasonally favorable Q4 effect in Brazil. Management guided continued portfolio diversification into lower-risk collateralized products (payroll loans, home equity) as the next phase of credit quality improvement.
Balance Sheet
Nu’s balance sheet is a function of two overlapping realities: a bank’s funded balance sheet (loans funded by deposits and equity) and a technology platform’s light operating asset base. Total assets are predominantly the credit card receivables, personal loan portfolios, and investment securities held to fund the deposit base. The deposit franchise — $41.9 billion in Q4 2025 — is both the most strategically important liability and the most operationally significant competitive advantage: it provides Nu with a low-cost, reliable funding source that is growing without requiring wholesale market access. The loan-to-deposit ratio has been improving as the deposit franchise scales faster than the credit book, implying Nu progressively self-funds its lending rather than relying on capital markets. Goodwill is minimal — Nu has grown almost entirely organically, with only the Easynvest acquisition adding meaningful acquired intangibles. This means Nu’s book value represents genuine tangible equity, not accounting goodwill inflated by acquisitions. Tangible book value per share is growing rapidly as the record-high net income compounds directly into equity, without material share dilution (management has been prudent on equity issuance). Capital ratios are well above minimum requirements, with the Brazilian regulators’ advanced IRB framework allowing Nu to hold capital commensurate with its actual credit risk profile rather than standardized risk-weight floors. No meaningful off-balance-sheet exposure exists. The balance sheet is structurally clean for a financial institution of this growth rate.
Cash Flow Statement
A bank’s cash flow statement requires the same analytical reframing applied to SoFi: the funded growth of the loan book consumes cash operationally, which is the correct deployment of a bank’s capital. The more instructive metrics are: (1) return on equity (33% in Q4 2025 — exceptional), (2) net income to equity contribution (Nu’s retained earnings are growing equity faster than almost any financial institution globally), and (3) cost efficiency ratio (19.9% — implying $0.80 of every revenue dollar is retained after non-interest expenses, before provisioning). Underlying technology infrastructure CapEx is modest relative to revenue — Nu does not own branches, ATMs, or physical distribution networks. The primary capital deployment is provisioning for expected credit losses on the growing loan book, which is both an income statement expense and a balance sheet reserve. Nu’s provisioning discipline is central to the investment thesis: under-provisioning would temporarily inflate net income while creating future credit problems; over-provisioning would depress reported earnings and TBV; management’s track record — 33% ROE and sequential improvement in NPL ratios simultaneously — suggests the provisioning framework is appropriately calibrated. Cash generation from the core Brazilian operation is substantial and compounding; the Mexico and Colombia operations are currently net cash consumers (early-stage investment) but are expected to reach Brazilian-level unit economics within 3–5 years, representing a step-function increase in group-level cash generation when that transition occurs.
The ARPAC Flywheel
The single most important KPI is Monthly Average Revenue per Active Customer (ARPAC): $10.7 in Q4 2024, rising to $13.4 by Q3 2025 , a 25% increase in roughly three quarters. More revealing: mature Brazilian cohorts from 2017-2018 already generate $25/month ARPAC, approximately 2.3x the current consolidated average.
This implies the ‘true’ revenue capacity of Nu’s current customer base is roughly double what it currently reports. The most mature cohorts (onboarded 2017-2018) are currently generating upward of $27/month ARPAC and rising as new products are introduced. As of Q1 2026, the blended $15.90 ARPAC is still less than 60% of where early cohorts now sit, indicating a multi-year compounding runway from product cross-sell alone.
The LTV/CAC ratio exceeds 30x. Nu’s blended CAC is approximately $5-7 per customer, of which only ~$2 is paid marketing. By the second quarter of a customer relationship, the CAC fully falls away, leaving only the $0.80-0.90 monthly cost to serve against an expanding ARPAC. Management formally estimates LTV/CAC at greater than 30x, a ratio that venture capital frameworks typically associate with elite SaaS businesses, not banks. The payback period is measured in months, not years.
Cost to serve (fixed)
$0.80
per customer per month
ARPU at full cross-sell
~$18
est. at max attachment
Margin at full ARPU
~96%
incremental revenue less cost
Estimated monthly ARPU by product attachment level vs. $0.80 fixed cost. The gap compounds with every product cross-sold.
Three Reinforcing Moats
Data Network Effect (STRENGTHENING): Nu’s underwriting uses 30,000+ data points per user accumulated over the customer’s life on the platform. Incumbent banks can match this for their own customers but cannot replicate it for the unbanked/underbanked entering credit for the first time , Nu often holds 3-5 years of behavioral data before a customer qualifies for a loan at Itau. Credit metrics from the first quarter of 2026 demonstrate the resilience of the underwriting model across broader economic cycles. The ninety day plus non performing loan ratio eased to 6.5% while the early stage delinquency ratio expanded to 5.0% reflecting an intentional strategic push into higher yielding customer cohorts. This sustained asset quality highlights the capability of the proprietary credit engine to price risk appropriately and maintain structural net interest margins even as the portfolio risk mix shifts. Quantified advantage: risk-adjusted NIM of 9.5% vs. ~5-6% for incumbents , a 350-400bps persistent spread.
Switching Costs (STRENGTHENING): A Brazilian consumer using Nu as a primary bank faces ~8-12 hours of administrative effort to transfer direct deposits and automatic payments, loss of accumulated credit history that took 2-4 years to build, and forfeiture of cashback rewards and investment products (Ultravioleta customers have $15,000+ in average AUM). These costs increase as product depth increases. Each additional product adds switching friction. Estimated switching cost: $400-600 in time/opportunity cost equivalent, up from ~$150-200 in 2021 when Nu was credit-card-only.
Cost Structure Moat (SUSTAINABLE): At $0.80/month cost to serve vs. R$30-60/month for incumbent branches ($6-12/month equivalent), Nu has a structural cost advantage of approximately 7-15x per customer. This is not temporary , it reflects the permanent absence of branch infrastructure and a 100% cloud-native tech stack. Nu’s marginal cost of serving the next customer approaches zero. This creates a permanent pricing floor below which incumbents literally cannot profitably compete.
Scale Economies Shared: Cultivating the Network Flywheel
A common failure mode for high-growth consumer enterprises is the temptation to extract maximum near-term value from their user base once market dominance is achieved. Nu has explicitly rejected this path. Rather than widening spreads as its cost advantages compound, the company has strategically passed operating efficiencies back to the customer in the form of zero-fee accounts, superior deposit rates, and lower credit costs than incumbents. This deliberate restraint is not altruism , it is the compounding mechanism itself. By offering demonstrably better economics, Nu sustains the organic word-of-mouth acquisition engine that keeps its blended CAC below $7. As each new customer validates the superior product, they become an unpaid advocate, compounding the customer base at near-zero marginal acquisition cost. This dynamic , sharing the scale economics rather than capturing them entirely , creates a self-reinforcing flywheel that is structurally difficult for incumbents to replicate: matching Nu’s pricing would require them to cannibalize the high-fee income streams that fund their physical branch networks.
NuFormer: AI as Competitive Infrastructure
Nu’s Q3 2025 strategic update explicitly positioned the company as ‘AI-first’, unveiling NuFormer , a proprietary foundational AI model trained on the company’s 30,000+ behavioral data points per customer. NuFormer is not a customer-service chatbot. It is applied directly to credit underwriting, enabling Nu to accurately assess creditworthiness for populations that traditional credit-score models classify as ‘invisible’ or ‘unbankable’. This capability has two compounding effects. First, it expands Nu’s effective TAM beyond the formally employed, formally banked population , a critical advantage in markets like Mexico and Colombia where large swaths of the adult population lack formal credit history. Second, it reduces NPL ratios over time as the models mature, a pattern already visible in Brazil where 15-90 day NPLs declined from 4.4% to 4.1% despite 57% loan portfolio growth. The proprietary financial behavioral dataset Nu has accumulated over more than a decade in Brazil represents arguably the most valuable credit intelligence asset in Latin America, and it cannot be purchased or replicated by any competitor starting from scratch today.
The Growth Story: Three Distinct Engines
Existing customer expansion (NRR): As Brazil cohorts mature from $10.7 to $25 ARPAC over 5-7 years, existing customers generate ~10-12% annual revenue growth from the base alone. Net Revenue Retention on the Brazil base is estimated at 115-125% annually , the safest and highest-quality revenue growth.
Geographic expansion: Mexico at 91% YoY customer growth in 2024. With 100M+ adults and only 10M customers, Mexico’s TAM runway is a decade long. The April 2025 banking license unlocks payroll accounts, higher deposit limits, and SME products , estimated $2B incremental annual revenue potential by 2027.
The geographic total addressable market expanded significantly following the recent announcement of a measured entry into the United States banking sector. Management established strict financial parameters for this endeavor and capped the investment impact at a maximum of one hundred basis points on the consolidated efficiency ratio through 2027. This disciplined approach provides an asymmetric growth option in a highly lucrative market without jeopardizing the broader profitability inflection and margin expansion trajectory modeled for the core Latin American business.
New products (option value): NuCel (MVNO), NuTravel, secured lending (growing 615% YoY from a small base), Ultravioleta (high-income segment at $1.8B quarterly purchase volume). Plus AI: Nu’s Q3 2025 release explicitly positions the company as ‘AI-first,’ with 30,000+ behavioral data points per customer representing arguably the most valuable proprietary financial dataset in Latin America.
The Profitability Inflection
The data tells a precise story: Nu moved from -$364M net income (2022) to +$1.03B (2023) to +$1.97B (2024) , nearly doubling net income YoY in each of the last two years. FCF margin expanded from 21.6% (2022) to 26.9% (2024).
Operating leverage is exceptional: Nu’s efficiency ratio of 29.9% in Q4 2024 compares to 35-40% for best-in-class global digital banks and 50-65% for Brazilian incumbents. The Mexico/Colombia combined operating loss (~$300-400M/year) creates an artificial ceiling on reported margins. When these markets inflect to profitability (modeled at 2027-2028), the efficiency ratio drops 200-400bps in a single year. Incremental operating margin on the next $1B of revenue: approximately 55-65% vs. blended current margin of ~30-35%.
Management validated the strength of the operating model in early June 2026 by authorizing a one billion dollar share repurchase program. That decision underscores board confidence in the underlying free cash flow generation and operating leverage of the digital platform.
Where the Market May Be Wrong
Error 1: Linear margin models. Sell-side models universally apply linear margin expansion. This misses the step-function dynamic from Mexico/Colombia inflection. Analysts are underestimating the profitability step-change by 12-24 months.
Error 2: ARPAC expansion underappreciated. The 2017-cohort data point , $25/month ARPAC , is treated as a theoretical ceiling rather than a demonstrated floor. At 85M active Brazilian customers averaging $20/month by 2029, Brazil alone generates $20.4B annually , vs. analysts modeling $14-16B.
Error 3: FX noise masks real business strength. The gap between $8.27B (USD-reported 2024) and $11.5B (FX-neutral 2024) creates persistent confusion. Most consensus models use USD-reported only, which understates local-currency operating momentum when BRL is weak.
The central analytical mismatch: the market often treats Nu as a mature bank when it is still a developing compounder with the financial profile of a mature one.
Key Risks
Brazilian credit cycle (35% probability): If unemployment rises 300+ bps, NPL 90+ ratio could spike from 7% to 11-13%, requiring $2.5-3.5B in additional provisioning and eliminating 2 years of net income.
BRL/MXN/COP depreciation (60% probability): Structural EM FX weakness erodes reported USD revenue even as local-currency business grows strongly. The gap between FXN and reported figures was 39% in 2024.
Regulatory capital requirements (30% probability): BACEN increasing minimum capital ratios would constrain lending growth and compress ROE. The former BACEN president joining as Vice Chairman (July 2025) partially mitigates this risk.
Mexico margin destruction (35% probability): Aggressive deposit pricing competition from BBVA/Banorte could sustain losses beyond 2027, impaired $1-2B in projected 5-year earnings contribution.
Prepared by Consti Ertel | March 1, 2026 | ConstiErtel.com | Not investment advice.