NU Nu's Margin Just Compressed on Funding Costs, Not Credit Risk. That's a Different Story.

Funding Costs, Not Credit Losses, Are the New Margin Story

Three and a half years after the IPO chronicled in nu/2021-12 and nu/2022-03, Nu Holdings' second quarter of 2025 - three months ended June 30, 2025 - looks, on the surface, like the same flywheel story: revenue hit a record $3.67 billion (+28.8% nominal YoY, +40% on Nu's preferred FX-neutral^» basis - the Brazilian real weakened against the dollar this quarter, the more typical direction for Nu's reporting, unlike the brief 2022 real appreciation flagged in the prior post), and net income hit a record $637.0 million, up 30.7% nominal YoY. The customer base crossed 122.7 million (+17% YoY), and monthly ARPAC^» broke $12 for the first time.

But underneath the headline growth, gross margin fell from 47.7% a year ago to 42.2% this quarter - a 550-basis-point compression, and not primarily a credit story this time. Total cost of financial and transactional services grew 42.4% (faster than revenue's 28.8%), but the fastest-growing line inside it wasn't credit loss allowance expense (+33.2% YoY, in line with the loan book's growth) - it was interest and other financial expenses, up 54.8% YoY to $1.03 billion, as Nu's deposit base (+41% YoY FXN to $36.6 billion) and its Central Bank funding costs both grew faster than the yield Nu earns on its own lending. That's a funding-cost squeeze, not a credit-quality one - a different, more structural pressure than the underwriting-driven compression flagged in nu/2022-03, and one that tracks with Brazil's own high domestic interest-rate environment rather than anything specific to Nu's risk appetite.

The Prescription

Nu should keep leaning into what's actually working here: converting a nearly-123-million-customer base into more products per customer while keeping cost to serve flat. Monthly cost to serve per active customer held at $0.80 this quarter even as ARPAC crossed $12 - the gap between those two numbers is the entire operating-leverage thesis, and it's still widening. Nu should keep pushing multi-product penetration (55 million active credit customers, 36.2 million active investment customers, 6.6 million crypto customers) rather than chasing pure customer-count growth, since the monetization curve on existing cohorts is still the more reliable lever than net adds in a market where Nu is already the third-largest financial institution in Brazil by customer count.

What Nu should stop doing: leaning on FX-neutral framing to describe the business's momentum without giving equal billing to what the nominal (as-reported, USD) numbers show. Every headline metric in the Q2'25 release is quoted in FX-neutral terms first, and the 40% FXN revenue growth number reads very differently from the 28.8% nominal number a reader would compute from the filed statements themselves - and margin compression, which is a purely nominal, currency-agnostic result (both sides of the ratio are in the same currency), gets no FXN framing at all because there's no favorable way to present it. A company confident in its FX-neutral story should be equally willing to show the reader the plain nominal numbers side by side, not lead with the flattering measure and let margin compression pass with less discussion than the growth metrics get.

Key Financial Metrics

Q2 2025 vs. Q2 2024, consolidated, reported in USD - Nu reports natively in US dollars, so no FX conversion is needed

Metric	Q2 2025	Q2 2024	YoY
Total revenue	$3,668.5M	$2,848.7M	✅ +28.8% nominal (+40% FXN)
Gross profit	$1,548.0M	$1,359.4M	⚠️ +13.9% (margin: 42.2% vs. 47.7% - see note above)
Total operating (expenses) income	$(667.6)M	$(634.0)M	✅ +5.3%, well below revenue growth
Profit before income taxes	$879.4M	$725.4M	✅ +21.2%
Net income for the period	$637.0M	$487.3M	✅ +30.7% nominal (+42% FXN)
Adjusted Net Income, non-IFRS	$694.5M	n/a (not separately disclosed for Q2'24 in this filing)	—
Total assets (period-end)	$62,730.1M	—	✅ +25.6% vs. Dec 2024 ($49,931.2M)
Total equity (period-end)	$9,576.1M	—	✅ +25.2% vs. Dec 2024
Deposits (period-end)	$36,640.2M	—	✅ +27.0% vs. Dec 2024 (+41% YoY FXN per company)

Annualized ROE^» was 28% for the quarter, per management. Operating expenses grew only 5.3% YoY against 28.8% revenue growth - genuine operating leverage below the gross-profit line, even while gross margin itself compressed on the funding-cost dynamic described above.

Key Operational Metrics

• Customers: 122.7 million globally (+17% YoY, +4.1 million net adds this quarter). Brazil: 107.3 million (over 60% of Brazil's adult population); Mexico: 12 million (~13% of Mexico's adult population); Colombia: 3.4 million (~10% of Colombia's adult population, deposits +841% FXN YoY off a small base).
• Monthly ARPAC: $12.2, up 18% YoY FXN - crossed $12 for the first time. Customers on the platform 8+ years monetize at $27.3/month, more than double the blended figure - the same cohort-maturation story flagged since nu/2022-03.
• Monthly cost to serve per active customer: $0.80, stable YoY.
• Asset quality: the 15-90 day NPL^» ratio (Brazil only) fell 30bps QoQ to 4.4%; the 90+ NPL ratio rose 10bps to 6.6%, which Nu attributes to normal seasonal follow-through from Q1's usual delinquency uptick rather than a new deterioration.
• Credit portfolio: total receivables across credit cards and loans reached $27.3 billion (+40% YoY FXN, +8% QoQ FXN); the Interest-Earning Portfolio (the subset actually generating net interest income) grew 55% YoY FXN to $15.7 billion. Net interest income hit an all-time high of $2.1 billion (+33% YoY FXN), with net interest margin up 20bps to 17.7% and risk-adjusted NIM (net of expected credit losses) up 100bps to 9.2% - the risk-adjusted figure improving even as the raw NIM only inched up is itself a sign credit quality, not funding cost, is the more favorable trend this quarter.
• Multi-product engagement: 55 million active credit customers (+11% YoY); 6.6 million Mexico credit cards (+52% YoY) and 1.4 million Colombia credit cards (+34% YoY); 36.2 million active investment customers (+70% YoY) and 6.6 million active crypto customers (+41% YoY).
• Segment/geography: still a single reportable operating segment (the CEO, as Chief Operating Decision Maker, reviews the business on a combined basis) - same structure as every prior Nu post. Country-level revenue is not disclosed in this interim filing (annual-only disclosure, consistent with nu/2022-03).

Beyond the Usual

A guaranteed, undrawn credit facility from a development finance institution doesn't show up in headline debt

Nu Holdings guarantees an undrawn credit facility that Nu Colombia holds with the U.S. International Development Finance Corporation (DFC), subject to covenants tied to capital adequacy, funding/liquidity position, and leverage ratios including net-debt-to-gross-profit - standard project-finance-style terms, but a parent guarantee on subsidiary borrowing that wouldn't appear in Nu Holdings' own consolidated debt figures unless drawn. This is the same category of cross-border parent guarantee flagged for Nu Servicios (Mexico) in nu/2022-03, now extended to the newer Colombian operation as it scales.

Nu's provision for lawsuits and administrative proceedings grew from $22.6 million at year-end 2024 to $27.9 million at quarter-end - a 23% increase in six months, driven almost entirely by civil lawsuits related to credit card operations (management describes these as arising "in the ordinary course of operations"). It's a small absolute number relative to a $9.6 billion total-liabilities balance sheet, but the growth rate is worth tracking as the credit card and lending books both keep expanding faster than the overall balance sheet.

Related-party transactions remain limited to routine items: executives, board members, and close family members can hold Nu credit cards, loans, deposits, and investment products, but only on the same terms offered to unrelated customers - not the kind of preferential-terms related-party lending that would be a genuine governance concern. Since 2019 (under Nu's predecessor private-company structure, well before the 2021 IPO), the company has repurchased roughly 16% of its own capital through an active buyback program funded alongside dividends and interest-on-capital distributions - a detail that doesn't appear anywhere in the quarterly press release's headline metrics but shows a long-running capital-return program operating well before Nu had positive net income as a public company.

Management Frames the Margin Story Sequentially, Not Year-Over-Year

CFO Guilherme Lago's prepared remarks described gross margin as "climbing to 42.2%, up from 40.6% in Q1" - a genuinely true and favorable sequential comparison - without once mentioning the year-over-year figure (47.7% in Q2'24) that shows the larger compression flagged in Key Financial Metrics above. Lago attributed the underlying quarter's dynamics mainly to a deliberate front-loading of credit loss provisions tied to a mid-quarter credit-model upgrade that widened credit limits in Brazil - a "temporary timing mismatch," in his words, between recognizing the provision now and collecting the corresponding interest income later - and separately flagged that investment in Mexico and Colombia "naturally weigh on short-term margins." Both explanations are plausible and consistent with the funding-cost dynamics in this post (see Beyond the Usual), but neither is the same claim as "margin improved," which is what the sequential framing implies to a listener who isn't cross-referencing the YoY numbers themselves. Founder and CEO David Vélez's opening remarks led with the 85% annualized revenue growth rate since 2021 and a 34% ARPAC CAGR - strong, real numbers, but ones that (like the margin discussion) favor the multi-year and sequential lenses over the harder year-over-year comparison this quarter's own margin trend would show.

Target Valuation Range

Reverse-DCF: at a $67.2 billion market cap, the market requires only ~20-22% sustained annual earnings growth for 7-8 years (against an 11-12% discount rate) - Nu's trailing net income growth has run 30-45% YoY, comfortably above what's priced in, though that gap makes a re-rating rather than an earnings miss the more likely source of disappointment from here.

Nu's stock has been genuinely volatile over the two years leading into this quarter: from a low near $6.85 (August 2023) it more than doubled to $14.97 by August 2024, then fell back to $10.24 by March 2025 - a roughly 32% drawdown - before recovering to close Q2 2025 (June 30, 2025) at $13.72. None of that swing tracks cleanly with the operating numbers shown above, which have grown steadily throughout; it reads more like a re-rating of the multiple the market is willing to pay for that growth than a reaction to any single quarter's results.

With approximately 4.90 billion diluted weighted-average shares outstanding:

Market cap → enterprise value	Q2 2025
Share price (period-end)	$13.72
Shares outstanding	~4.90B
Market capitalization	~$67.2B
Total liabilities (borrowings and financing)	~$2.3B
Less: cash and equivalents	~$13.3B
Enterprise value (net-cash basis)	~$67.2B (minimal net debt)

Trailing-twelve-month net income (Q3'24 through Q2'25, summing the parent-attributable figures disclosed across Nu's own filings: $553.4M + $552.6M + $557.2M + $637.0M) comes to approximately $2,300.2 million.

Peer-multiple sanity check	Q1 2025	Q2 2025	Change
Enterprise value	~$40.8B	~$67.2B	⚠️ up sharply
P/E (trailing)	~22.2x (annualized run-rate)	~29.2x (TTM)	⚠️ up

At ~29x trailing earnings, Nu trades at a large premium to Brazilian incumbent banks (Itaú Unibanco and Banco Bradesco typically trade 8-10x trailing earnings) and to other emerging-market digital banks, reflecting both Nu's much higher growth rate and its lower loan-loss-driven earnings volatility so far. It's a fair comparison only up to a point - incumbents have decades of seasoned loan books and full-cycle credit history that Nu's still-young Mexico and Colombia operations don't yet have.

Reverse DCF: at $67.2 billion enterprise-equivalent value (Nu carries minimal net debt - $2.3 billion of borrowings and financing against $13.3 billion of cash) and ~$2.3 billion of TTM net income, the market is pricing in roughly 20-22% sustained annual earnings growth for the next 7-8 years before terminating at a mature-bank multiple (~12-15x), using an 11-12% discount rate reasonable for a Brazil/Mexico/Colombia-exposed financial (higher than a developed-market bank's cost of equity given currency and regulatory risk). Nu's trailing net income growth has run 30-45% YoY the last several quarters - comfortably above what's priced in, which is the bull case in one sentence, but it also means multiple compression (a re-rating toward a lower growth assumption) rather than earnings misses is the more likely source of a stock-price disappointment from here.

The reverse-DCF and peer-multiple lenses above are the more honest read for now.

---