March 26, 2026
Emerging Markets Coverage · Initiation
Nu Holdings: The Mispriced Compounder
Conviction Long NU · NYSE · Base PT $18
Published by LIRIA RESEARCH · Independent Equity Research

The market has decided Nu Holdings is a Brazilian fintech. What it has not decided is how to price a business with 33% return on equity, $0.80 cost per customer per month, and 45% net income growth, operating without a single physical branch. At $14.17, the trailing P/E is 24x on earnings that grew 45% last year. This is an initiation at $14.17 with an $18 base case price target and a $26 bull case.

The geographic discount is real. The question is whether it is rational when applied to a business that has better unit economics than any large bank in the Western world.

62% of Brazil's adult population now has a Nu account. This is not a challenger bank disrupting the edges of the market. It is the market.
What Nu Actually Is

Nu Holdings is a financial services company founded in São Paulo in 2013. It launched with a single product: a no-fee credit card managed entirely through a mobile app, targeting the 50% of Brazilian adults who were unbanked or underbanked and paying extortionate fees to legacy institutions.

Thirteen years later, it is the largest private financial institution in Brazil by customer count, with 131 million customers across Brazil, Mexico, and Colombia. The product suite now covers credit cards, checking accounts, personal loans, high-yield savings, investments, insurance, and SME banking. The company runs all of it without a single physical branch.

The founding insight that still drives everything: legacy Brazilian banks were not expensive because banking was inherently expensive. They were expensive because they had built an infrastructure of branches, paper processes, and manual underwriting that required enormous headcount to maintain. Remove the infrastructure, rebuild underwriting with data, and the cost structure collapses.

That is what happened. The result is $0.80 per customer per month. Itaú Unibanco, Brazil's largest bank, operates at roughly $15 to $20 per customer per month. Nu serves 131 million people at a fraction of that cost. At scale, this is not a temporary advantage. It is structural.

The Unit Economics — Why This Business Is Different

The most important number in the Nu thesis is not the customer count. It is not even the revenue. It is the spread between the $0.80 monthly cost and the $15 monthly revenue per active customer. That spread, multiplying across a growing base and expanding ARPAC, is the compounding engine.

ROE Q4 2025
33%
JPMorgan: ~16%
Cost / Customer / Mo
$0.80
Legacy peers: $15-20
NI Growth FY2025
+45%
On +27% revenue
Revenue FY2025
$6.99B
+26.8% YoY
Net Income FY2025
$2.87B
+45.5% YoY
ARPAC Q4 2025
$15
+27% YoY

The 45% net income growth on 27% revenue growth is the critical observation. Every incremental dollar of revenue is converting at a higher rate than the last. This is operating leverage in a business with near-zero marginal cost per customer. Traditional banks cannot produce this because their cost base scales with their customer base. Nu's does not.

The $0.80 figure deserves to be taken seriously rather than treated as a marketing number. It reflects the fundamental economics of zero-branch banking plus AI-driven underwriting. There are no physical assets to maintain, no teller wages, no lease obligations on 10,000 branches. The underwriting model runs on transaction data from 131 million accounts. Traditional lenders cannot replicate this without dismantling their existing infrastructure, which is economically and operationally impossible in the near term.

Net Income Trajectory — FY2022 to FY2025
From loss to $2.87B in three years · Hover bars for detail
$3B $2B $1B $0 -$0.5B -$0.4B FY2022 $1.0B FY2023 $1.97B FY2024 $2.87B FY2025
Source: Nu Holdings earnings reports. FY2023 was the first full profitable year.

The NPL picture is also moving in the right direction. The 90-day NPL rate declined from 5.6% in 2024 to 4.4% in 2025 while the credit book grew significantly. Improving credit quality during rapid credit growth is unusual and reflects the proprietary advantage of underwriting from 15 years of transaction data across the entire base.

The ARPU Engine — Revenue Without New Customers

The customer growth story in Brazil is largely over. 131 million accounts covering 62% of the adult population is saturation, not runway. This is often cited as a bear case. It is actually the wrong way to frame the business.

The revenue growth story in Brazil is about ARPU expansion, and it has barely started. The average active Nu customer generates $15 per month in revenue today, up 27% from a year ago. A customer with a checking account, a credit card, personal loans, an investment account, and insurance products generates $35 to $50 per month. Most Nu customers have two or three products. Almost none have all of them.

Each product launch follows the same pattern. The customer already trusts Nu with their primary account. The friction to add another product is low. The incremental acquisition cost is zero. The data advantage for pricing risk is higher with every additional data point. This is why ARPU compounds even when customer count plateaus.

Monthly ARPU per Active Customer — Q4 2022 to Q4 2025
Compounding without acquiring new customers · Hover dots for detail
$16 $12 $8 $4 Q4 2022 Q4 2023 Q4 2024 Q4 2025 $7 $10 $11.8 $15
Source: Nu Holdings quarterly earnings. ARPU per active customer, USD.

The path to $25 per month ARPU is not speculative. NuInvest (the investment platform) has over $8B in AUM and is growing. NuSeguros (insurance) is early. SME banking, launched for Brazil's enormous MEI (individual microentrepreneur) segment, has barely begun. Each of these expands the revenue per existing customer without requiring a new account to be opened.

The compounding is straightforward: if 131 million customers move from $15 to $22 in monthly ARPU over the next three years, that is $60B in additional annual revenue from the existing base alone, before Mexico or Colombia contribute a single dollar.

Mexico — The Second Act

Mexico is 85 million adults, a banking penetration rate of roughly 49%, and a credit card penetration far below Brazil's. Nu entered Mexico in 2022 and now has approximately 8 million customers there. The credit card was the entry product. Personal loans and a full digital account are rolling out.

The bear case on Mexico is that thin credit bureau data makes underwriting harder and NPL risk higher. This is true, but the conclusion drawn from it is backwards. Thinner credit bureau data does not disadvantage Nu. It disadvantages legacy lenders who depend on bureau data precisely because they have no proprietary transaction history. Nu is building that proprietary data set in real time, from day one of every Mexico customer relationship.

The Brazil playbook took nine years to cover 62% of the adult population. Mexico, starting from a lower banking penetration base and with a proven model, is tracking faster. The unit economics argument that worked in Brazil applies with equal force in Mexico. Legacy Mexican banks operate with the same branch-heavy cost structure. Nu does not.

Nu Competitive Moat vs Traditional Banks
The cost structure advantage persists at every market stage
Metric Nu Holdings Itaú (Brazil) Banorte (Mexico)
Cost / Customer / Month $0.80 ~$18 ~$15
ROE 33% ~20% ~18%
Physical Branches 0 ~4,300 ~1,300
Credit Underwriting AI, proprietary data Bureau + credit history Bureau + credit history
Net NPA 4.4% (improving) ~2.9% ~1.8%
Source: Company reports, FY2025. Nu's higher NPL reflects a younger, previously unbanked customer base, not underwriting failure.

Colombia is smaller and earlier. It is worth watching as an optionality call, not a near-term driver. The Mexico opportunity alone, at Brazilian-scale penetration, would add tens of millions of customers to the base over the next decade.

The Setup — Why the Mispricing Exists and Why It Matters Now

Nu trades at 24x trailing earnings on 45% net income growth — a PEG ratio of approximately 0.53. The honest anchor is the earnings trajectory itself, not a cross-sector multiple comparison. Comparing Nu's P/E to JPMorgan or PayPal is a category error: those multiples reflect single-digit growth, mature revenue mix, and completely different cost structures built around trading desks, investment banking, and physical infrastructure. A business compounding net income at 45% annually should command a growth premium. At 24x, the market is paying a distressed-growth multiple for a compounding machine.

Why does this discount exist? Three reasons, all geography-based.

First, institutional EM mandates are structurally underweight Brazil. Most global EM funds benchmark against the MSCI EM index, where Brazil has shrunk as a weight over the past decade. The analysts covering EM banks are generalists who price Brazilian risk using macro overlays, not bottom-up unit economics. Nu gets lumped into "LatAm fintech" and discounted accordingly.

Second, BRL volatility creates headline risk. When the Brazilian real depreciates, Nu's USD-reported revenue and earnings fall even when the underlying business is performing. This creates periodic selling pressure from investors who do not distinguish between BRL business performance and USD reporting. The fundamental business can be accelerating while the stock falls.

Third, the comparison set is wrong. Nu is consistently compared to neobanks like Revolut, Monzo, and Chime, which are subscale and largely unprofitable. Against any efficiency benchmark — ROE, cost-to-serve, ARPU growth — Nu has no real peer. That is the anomaly the market is mispricing.

There is a steelman to the discount that deserves engagement rather than dismissal. The market is not simply being lazy when it applies an EM overlay to Nu. It is pricing a legitimately higher cost of equity for a business domiciled in an economy where currency crises, inflation spikes, and populist regulatory interventions can materially impair a lender's compounding engine in ways that have no real analogue in developed markets. Brazil's equity risk premium — typically 300–400 basis points above US — is not irrational. It reflects a macroeconomic history that includes multiple currency collapses, credit freezes, and regulatory reshufflings. The counterargument here is not that the EM discount should be zero. It is that the current discount is excessive: the market is embedding a 2015-era Brazil risk overlay onto a 2025-era business that has already demonstrated it can compound through exactly the kind of environment the bears describe.

Net income growing at 45% per year, at a 24x trailing multiple, implies the market expects this growth to disappear within two years. The underlying business does not support this expectation.

The reason now is that the compounding has become undeniable. Three consecutive years of profitability, a declining NPL rate during rapid credit growth, and ARPU expansion driven by products that barely existed 18 months ago. The narrative risk that dominated 2022 and 2023 (will this business ever be profitable?) has been resolved. What remains is the valuation gap between what the business is and what the market is paying for it.

Scenarios
Price Target Scenarios
Initiation at $14.17 · Base PT $18 · Bull PT $26 · Bear PT $9
Scenario FY2027E NI Multiple Price Target Return
Bull — ARPU reaches $22, Mexico accelerates $5.5B 23x $26 +82%
Base — ARPU steady, Mexico gradual $4.2B 21x $18 +26%
Bear — BRL -25%, NPL spike in Mexico $2.8B 16x $9 -37%
Assumes ~4.85B shares outstanding. Base case uses 21x forward P/E, modest for 30%+ earnings growth. Brazil's country risk premium (~300–400bps above the US) is already embedded in the discount rate and in the stress-case earnings assumptions — applying an additional blanket multiple haircut on top of that would be double-counting the same risk.

The bull case does not require extraordinary assumptions. It requires ARPU to grow from $15 to $22 over two years — a pace consistent with the historical trajectory — and Mexico to continue on its current path. No step-change in the business is needed.

The bear case is driven almost entirely by macro factors: a severe BRL depreciation combined with credit quality deterioration in new markets. This is the legitimate tail risk, and it is why position sizing matters. The business itself is not the risk. The currency exposure and the new-market credit book are the risks.

Key Risks
Liria Research · Verdict
Conviction Long — Initiation at $14.17

The market is applying an emerging market discount to a business that has never been about emerging markets. It has been about building the most efficient financial institution ever constructed, in a geography where the incumbents were the most inefficient. That strategy has produced 33% ROE, $0.80 per customer per month, and $2.87B in net income growing at 45% per year.

At 24x trailing earnings on 45% net income growth, the market is paying a multiple that implies growth disappears imminently. The ARPU expansion, the Mexico trajectory, and the product roadmap suggest it does not. The base case of $18 requires no extraordinary assumptions. The bull case of $26 requires only that the business continues doing what it has done for three years.

BRL exposure is the real risk — not merely as a reporting artifact, but as a macro mechanism that correlates currency stress with NPL deterioration and regulatory intervention simultaneously. The EM discount is partially rational: Brazil's equity risk premium reflects a genuine history of macro volatility that developed-market multiples do not contemplate. What is irrational is the magnitude of the current discount, which implies either imminent business failure or permanent macro dysfunction — neither of which is supported by three consecutive years of compounding through exactly the kind of environment the bears describe. Size accordingly.