Financials

Financials — Nu Holdings Ltd. (NU)

1. Financials in One Page

Nu Holdings is a Brazil-headquartered digital bank running on a lender's chassis: it earns most of its revenue from net interest income (interest on credit cards plus consumer loans, minus interest paid on deposits) and the rest from fees on cards, FX, marketplace, and investments. Top-line growth has been violent — total revenue (NII + fees, before provisions for credit losses) compounded from $0.27B in FY2018 to $11.2B in FY2025, a 70% five-year revenue CAGR — and the business crossed into GAAP profit in FY2023. FY2025 delivered $2.87B of net income at a 41% net margin on net revenue (after credit provisions), 30.3% ROE, and $3.49B of free cash flow with FCF margin above 50% of net revenue. The balance sheet now carries $74.9B of assets, $41.9B of deposits, and $27.7B of loans — funded almost entirely by customer deposits at 6.5x assets-to-equity, well inside Brazilian incumbent leverage. The fight is no longer whether Nu is real; it is whether the market is paying the right premium for it. Q1 2026 results (May 14) — $5.0B revenue, $871M net income, +41% net income YoY but a 1¢ EPS miss — knocked the stock from a $18.76 52-week high to $12.07 on May 15, a 36% drawdown that resets the valuation question. The metric that matters most right now: provisions for credit losses as a percentage of net interest income, because Nu's earnings growth from here depends on whether credit costs stabilize as the Mexican loan book ramps.

Definitions used on this page. Gross Revenue = Net Interest Income + Non-Interest Income (the top line of a bank, before credit costs). Net Revenue = Gross Revenue minus Provision for Credit Losses; the data provider uses this as the denominator in margin ratios, so we report both. Net Interest Income (NII) = interest earned on loans, cards, and securities minus interest paid on deposits and debt. Cost of Risk = Provision for credit losses divided by average net loans; for a consumer lender this is the single most volatile line. Free Cash Flow (FCF) in this report = Operating Cash Flow minus Capex, exactly as the cash-flow file reports it. For a bank that is mostly an accounting-style cash measure dominated by deposit growth; do not read it the way you would for a software company.

2. Revenue, Margins, and Earnings Power

Nu's revenue engine is built on two layers. First, customer acquisition: from roughly 60M customers in FY2022 to ~131M at year-end 2025, with 14M in Mexico and 4.2M in Colombia. Second, monetization: average revenue per active customer rises as Nu cross-sells credit cards, then unsecured loans, then payroll loans, investments, and marketplace products. The result is revenue compounding faster than customers — NII compounded at ~85% over five years vs. asset growth of ~49% — because the mix is moving toward higher-yielding lending products.

Read the chart this way. Gross revenue (purple) is what Nu earned across NII and fees. Net revenue (blue) is what Nu kept after setting aside expected credit losses. The gap between purple and blue is the credit-cost line, and it has widened from $119M in FY2018 to $4.2B in FY2025 as the loan book grew almost 18x. NII (yellow) became the dominant component from FY2022 onward — Nu shifted from a fee-led card company to a true spread lender. Fee income (green) is now the smaller half of revenue but it carries very different economics (no credit risk, no funding need) and is a structural quality lever.

The margin slope tells the story neoplatform skeptics need to see. FY2018-FY2022: pre-tax margin pinned between -17% and -43% as Nu spent on customer acquisition and credit ramp. FY2023 is the inflection — pre-tax margin jumps to +41.5% as scale absorbs SG&A and as the loan book matures into a pricing-power asset. FY2025 pre-tax margin of 55.3% and net margin of 41.1% are not promotional; they are now structurally higher than every direct peer in the table on Section 8.

Two things to read here. Revenue trajectory remains clean: Q1 2026 hit a fresh $3.7B record, +57% YoY. Net income stalled sequentially in Q1 2026 (-2.6% QoQ to $871M) for the first time since Q4 2022, even as revenue accelerated, because credit provisions jumped 30%+ sequentially. This is why the EPS missed by 1¢ and the stock got cut. Cross-reference with the cost-of-risk view in Section 4.

3. Cash Flow and Earnings Quality

Free Cash Flow definition for this section. Operating Cash Flow minus Capex (capital expenditures on property, equipment, software). For a bank, operating cash flow is dominated by changes in operating assets/liabilities — primarily deposit inflows and loan funding. So Nu's reported FCF is not the same conceptual quantity as a software company's FCF; it is best read as "cash retained inside the business after funding loan growth and capex."

The cash story is unambiguously high quality once you understand the bank lens. For FY2023-FY2025, OCF runs at ~1.22x net income — meaning every $1 of GAAP earnings is matched by $1.22 of operating cash flow. The "extra" $0.20-$0.30 per dollar of earnings comes mostly from a fast-growing deposit franchise: customers leave money in NuAccount, which shows up as a cash inflow on the operating-activities section the bank classifies it under. SBC at $272M in FY2025 is now only ~3.9% of net revenue, down from a 33% peak in FY2022 when post-IPO equity awards weighed heavily.

The negative FY2021 OCF (-$2.9B) is the only red flag, and it has a clean explanation: 2021 was the IPO year, with credit-card receivables expanding at IPO-funded scale before deposits caught up. Once deposits scaled, the working-capital drag flipped to a tailwind.

4. Balance Sheet and Financial Resilience

For a digital bank, the right resilience frame is not "net debt / EBITDA" — that ratio looks alarming on banks because deposits are technically liabilities. The right frame is: capital adequacy (equity vs. risk-weighted assets), liquidity (cash + securities vs. deposits), credit quality (provisions vs. loans), and funding mix (deposits vs. wholesale debt).

Three takeaways from the balance sheet. First, deposits ($41.9B) > loans ($27.7B). Nu is over-funded — it has $14B more in customer deposits than it has lent out — and that gap is itself a moat: it depresses the cost of funds and gives Nu pricing latitude that pure-play lenders don't have. Second, cash + securities ($40.9B) > total loans ($27.7B). Liquidity is enormous; this is a bank that could survive a multi-quarter deposit run without selling impaired loans. Third, long-term debt jumped 2.5x in FY2025 to $4.4B. This is not distress; it is opportunistic local-currency funding to lengthen Nu's funding tenor as the loan book grows. Long-term-debt-to-equity is still only 0.39x.

The single most important risk metric for Nu is cost of risk — how much it has to set aside against bad loans, expressed as a percent of the average loan book. It climbed from ~5% in FY2020 to ~19% in FY2024 as Nu's mix shifted toward unsecured personal loans (high-risk, high-margin). FY2025 it dipped slightly to 18.6%. A stabilized cost of risk would be very bullish — it implies Nu has reached the equilibrium between aggressive growth and credit underwriting. A re-acceleration above 20% in 2026 would be the single biggest threat to the earnings trajectory.

Leverage of 6.6x is far inside Itaú (assets-to-equity historically around 12-14x for the Brazilian incumbents) and Bradesco (similar). Loans-to-deposits at 0.66 is conservative for a consumer bank — many global peers run 0.85-1.0. Liquidity (cash + securities) covers 1.49x of all customer deposits. Resilience is not the worry. Credit cost is.

5. Returns, Reinvestment, and Capital Allocation

Returns on capital have walked through three phases. FY2018-FY2022: deeply negative ROE (-7% to -33%) — Nu was a venture-stage business burning equity to acquire and lend. FY2023: positive turn at 18.2% ROE. FY2024-FY2025: structural compounder territory at 28-30% ROE, converging on Itaú's ~21% incumbent ROE benchmark and quickly surpassing it.

Capital allocation is, in one phrase, reinvest in the loan book. Nu has not paid a dividend, has not bought back shares, and has consumed almost no capex (under $10M annually in FY2024 and FY2025). Cash and earnings are being plowed back into the loan portfolio, which compounded from $5.9B in FY2021 to $27.7B in FY2025. That is the right call given the ROE on incremental capital is north of 30%.

The IPO step-up in FY2021/2022 (1.6B → 4.7B shares) added dilution that has bottomed out. From FY2023 to FY2025 diluted shares rose only 1.0% per year on net — SBC dilution is now mostly offset by employee withholding and small stock issuances. Per-share economics finally compound. Diluted EPS: $0.21 (FY23) → $0.40 (FY24) → $0.58 (FY25). Tangible book value per share: $1.18 (FY23) → $1.41 (FY24) → $2.09 (FY25) — a 33% TBV/share CAGR.

6. Segment and Unit Economics

Granular segment financials are not separately disclosed in the data provider's files. Geographic disclosure in the 20-F and management commentary indicate the mix is roughly: Brazil ~85-90% of revenue and the great majority of profit; Mexico the growth engine (14M customers, ~10% of the customer base, smaller revenue share, profitability still building); Colombia early-stage (4.2M customers).

Unit economics that the public financials let us infer at the group level: revenue per customer is rising (ARPAC) from roughly $7/month at IPO to mid-$11/month by year-end 2025 as cross-sell into loans and investments deepens. Cost-to-serve per customer remained roughly flat in dollar terms despite double-digit local-currency inflation in Brazil — that is the operating leverage that drove the margin step-change in Section 2.

The right read on segment mix: Brazil is funding the build, Mexico is the optionality, Colombia is a small option premium. If Mexico can reach mid-teens revenue contribution at Brazil-like margins (still a 3-5 year arc), the consolidated revenue line has another doubling in it. If Mexico stalls or shows credit-cost blow-up, group ROE compresses without much offset.

7. Valuation and Market Expectations

Nu trades like a high-growth bank, not a typical bank. At year-end FY2025 (price $16.74), it carried a P/E of 28.6x, P/B of 7.2x, and P/Sales of 11.6x. Those are software-grade multiples on a bank P&L. The May 14-15, 2026 EPS miss compressed the multiple — at the May 15 close of $12.07 the trailing P/E falls to roughly 20.6x and P/B to ~5.2x.

The right valuation lens for Nu is P/B vs ROE — the universal yardstick for banks. A bank trading at 1x book at a 10% ROE is identical math to a bank trading at 3x book at a 30% ROE. On that math Nu's ~5-7x P/B against a 30% ROE prints a P/B-per-ROE-point of roughly 0.20-0.24, which is cheaper than the marquee high-ROE banks (ITUB at 2.13x book / 21% ROE = 0.10 per point, the structural compounder pricing) and much more expensive than the low-ROE incumbents (BBD at 0.55x / 13.8% ROE = 0.04). The P/B-vs-ROE judgement: Nu is being priced as a high-quality growth bank but not as a runaway over-valuation, particularly after the May 2026 reset.

Consensus is Buy at a $15.81 target (16 analysts), with a range of $8.10 (bear) to $22.00 (bull — Goldman Sachs reiterated $21). At the current $12.07, that consensus implies ~31% upside. The bear case implicitly prices a credit-cost blow-up in 2026. The base case prices stable cost-of-risk + Mexico contribution turning positive. The bull case prices Mexico becoming a second engine and ROE staying above 25%.

8. Peer Financial Comparison

NU plus five peers: INTR (Brazil digital-bank pure-play), ITUB and BBD (Brazilian incumbents), MELI (LatAm fintech ecosystem), PAGS (consumer digital bank + acquiring). FY2025 figures from the company's own reported financials.

The peer table sharpens the read in three ways. First, Nu is the only peer combining 25%+ revenue growth, 40%+ net margin, and 30%+ ROE — that triple is rare in any market and rarer in banking. INTR matches the growth but at half the ROE and a quarter of the margin. MELI matches the ROE but at one-sixth the net margin. ITUB matches the profitability but at one-tenth the growth. Second, Nu's premium to incumbents (P/B of 7.2 vs. ITUB at 2.1 and BBD at 0.55) reflects the growth gap; on P/B-per-point of ROE Nu is closer to MELI (a growth-priced ecosystem) than to a bank. Third, Nu's market cap ($81.3B at FY25 close, ~$58.6B at the May 15 reset) is now in the same tier as ITUB ($79B) — Nu is not pricing as the challenger anymore; it is pricing as a peer of Latin America's largest private bank.

9. What to Watch in the Financials

What the financials confirm. Nu has crossed from venture-stage cash-burner to genuinely high-quality compounder — 30% ROE, 41% net margin, $3.5B FCF on a software-style fixed-cost base, and a fortress liquidity position with deposits 1.5x cash+securities cover. Earnings convert to cash. Capital is being reinvested where the marginal ROE is highest. Per-share value is compounding (diluted EPS up 44% YoY in FY2025, TBV/share up 49%). The 20-F balance sheet supports far more leverage than Nu currently runs.

What the financials contradict. The "premium valuation always" narrative. After the May 14 EPS miss and the 30%+ peak-to-trough drawdown, multiples are no longer aggressive on a P/B-per-ROE basis. The other contradiction: the bear-case fear of insolvency risk. Cost of risk is high but cleanly funded by the spread; provisions absorb less than half of NII; equity has grown faster than loans for three years.

The first financial metric to watch is provisions for credit losses as a percentage of net interest income. It dropped from 51% (FY2022) to 50% (FY2023) to 47% (FY2024) to 47.5% (FY2025) — a steady, encouraging decline. If FY2026 prints below 45%, the structural earnings-growth story stays intact at a premium rate; if it pushes back toward 55%, the setup points to further multiple pressure and a re-rating of the LatAm digital-bank cohort.