Industry

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.