NU Nu Deliberately Grew Into Lower-Yield Assets in Q3. Its Risk-Adjusted Margin Still Went Up.

A Different Kind of Margin Story Than Last Quarter

Nu Holdings' third quarter of 2025 (three months ended September 30, 2025) extends the growth trajectory covered in nu/2025-06: revenue reached a record $4.17 billion (+41.8% nominal YoY, +39% FXN^»), net income hit a record $782.7 million (+41.4% nominal YoY), and the customer base crossed 127 million (+16% YoY). ARPAC^» broke $13 for the first time, and annualized ROE^» hit a record 31%.

But the margin story this quarter is genuinely different from Q2's. Net interest margin contracted 40 basis points sequentially, from 17.7% to 17.3% - on its face, a continuation of the funding-cost squeeze flagged in nu/2025-06. CFO Guilherme Lago's explanation on the call, pressed hard by Morgan Stanley's Jorge Kuri, is more nuanced: Nu deliberately grew faster into lower-risk, lower-yield assets this quarter (secured lending outpaced the rest of the portfolio, and even within credit cards, growth skewed toward lower-risk customers who carry lower yields "all else equal"), while funding costs in Brazil rose in dollar terms as the local SELIC rate stayed at 15% even as Nu's cost of deposits fell as a percentage of the interbank rate (91% to 89%). The two effects - cheaper relative funding, more expensive absolute funding because the underlying Brazilian rate itself is high, plus a mix shift toward safer assets - netted out to a lower raw NIM. The tell that this is a quality trade-off rather than a genuine deterioration: risk-adjusted NIM (net of expected credit losses) actually rose, from 9.2% to 9.9% - the clearest evidence yet that Nu is choosing to originate less risk per dollar of assets, not losing pricing power.

The Prescription

Nu should keep making exactly this trade: originating into lower-risk segments even when it compresses the raw NIM, as long as risk-adjusted NIM keeps expanding the way it did this quarter (9.2% to 9.9%). A bank that optimizes for raw yield over risk-adjusted return is the bank that blows up a credit cycle down the line - and Nu's own 15-90 day NPL^» ratio improving to 4.2% (down 20bps QoQ) while growing the portfolio 42% YoY FXN is the proof this isn't just a defensive retreat but genuine underwriting improvement, aided by the AI-driven credit modeling management referenced on the call.

What Nu should stop doing: declining to quantify a disclosed one-off benefit when directly asked. On the call, Kuri followed up asking management to quantify the impact of stronger-than-expected recoveries from a "second chance" reactivation program for previously-defaulted customers on this quarter's credit loss allowance expense - Lago acknowledged the program and its effect but explicitly declined to give a number ("we are not disclosing this one-off impact"). If the number is material enough to be worth naming as an explanation for beating expectations, it's material enough to quantify; naming a favorable one-off without sizing it lets a reader assume it was small without actually establishing that.

Key Financial Metrics

Q3 2025 vs. Q3 2024, consolidated, reported in USD - Nu reports natively in US dollars, so no FX conversion is needed

Metric	Q3 2025	Q3 2024	YoY
Total revenue	$4,172.7M	$2,943.2M	✅ +41.8% nominal (+39% FXN)
Gross profit	$1,814.5M	$1,348.6M	⚠️ +34.6% (margin: 43.5% vs. 45.8% - smaller compression than Q2's)
Total operating (expenses) income	$(697.1)M	$(624.8)M	✅ +11.6%, well below revenue growth
Profit before income taxes	$1,116.3M	$723.8M	✅ +54.2%
Net income for the period	$782.7M	$553.4M	✅ +41.4% nominal (record)
Adjusted Net Income, non-IFRS	$829M	n/a (not separately disclosed for Q3'24 in this filing)	—
Total assets (period-end)	$68,362.8M	—	✅ +36.9% vs. Dec 2024 ($49,931.2M)
Total equity (period-end)	$10,553.5M	—	✅ +38.0% vs. Dec 2024
Deposits (period-end)	$38,775.9M	—	✅ +34.4% vs. Dec 2024 (+34% YoY FXN per company)

Efficiency ratio (operating expenses as a share of gross profit, Nu's own preferred productivity metric) improved slightly to 27.7% from 28.3% in Q2, reversing the RSU-vesting-driven uptick flagged last quarter.

Key Operational Metrics

• Customers: 127 million globally (+16% YoY, +4.3 million net adds). Brazil: 110.1 million (over 60% of Brazil's adult population, activity rate above 85%, deposit base $30.4 billion); Mexico: 13.1 million (~14% of Mexico's adult population); Colombia: 3.8 million (~10% of Colombia's adult population).
• Monthly ARPAC: $13.4, up 20% YoY FXN.
• Monthly cost to serve per active customer: $0.90, still below $1.00.
• Asset quality: 15-90 day NPL ratio (Brazil only) fell 20bps QoQ to 4.2%, slightly below typical Q3 seasonality; 90+ NPL ratio rose 20bps to 6.8%, in line with expected seasonal patterns.
• Credit portfolio: total credit portfolio expanded 42% YoY FXN and 9% QoQ FXN to $30.4 billion; the Interest-Earning Portfolio grew 54% YoY FXN to $17.7 billion. Net interest income hit a new all-time high of $2.3 billion (+32% YoY FXN) despite the NIM compression described above, since portfolio growth outran the margin decline.
• Segment/geography: still one reportable operating segment. This interim filing's geographic revenue footnote shows Brazil at $2,883.9M, Mexico at $220.0M, and other countries at $58.0M of the geographically-attributed revenue for the quarter - Brazil is roughly 91% of that disclosed geographic split, Mexico about 7%, the rest under 2%.

Beyond the Usual

Management disclosed, but explicitly declined to quantify, a one-off benefit to this quarter's credit loss allowance expense from stronger-than-expected recoveries on a "second chance" program for previously-defaulted customers who have since paid down their debt (see The Prescription above). It's disclosed rather than hidden, but naming a favorable one-off while refusing to size it makes it harder for a reader to judge how much of this quarter's asset-quality improvement is structural (the credit-model upgrade, the mix shift toward lower-risk segments) versus a one-time recovery benefit that won't repeat.

Nu's provision for lawsuits and administrative proceedings grew further to $30.3 million (from $22.6 million at year-end 2024 and $27.9 million as of Q2'25) - a steady upward trend across three consecutive filings, still dominated by civil lawsuits tied to credit card operations and still small relative to the balance sheet, but worth tracking as a trend rather than a one-quarter blip.

Nu's related-party disclosures remain limited to routine, market-rate transactions with executives, directors, and close family members (credit cards, loans, deposits, investment products on the same terms offered to unrelated customers) - no preferential-terms lending or governance red flag identified this quarter, consistent with every prior Nu filing reviewed.

Management's Case for the Margin Trade-Off

CFO Guilherme Lago's response to Jorge Kuri's pointed NIM question (transcribed above under The Real Driver) is the single most substantive piece of disclosure in this quarter's call - a detailed, numbers-first walk-through of exactly why NIM fell in dollar terms even as the cost of deposits fell as a percentage of the interbank rate. That's a materially more transparent answer than the sequential-only margin framing flagged as this blog's critique of nu/2025-06 - this quarter, when a specific analyst pushed on the exact mechanics, management gave a real answer rather than reframing the question. The one place management didn't fully follow through on transparency is the recoveries-quantification question in Beyond the Usual above - a smaller gap, but the same instinct (favorable framing without the number that would let a reader size it) showing up again.

Target Valuation Range

Reverse-DCF: at a $78.5 billion market cap, the market requires ~20-23% sustained annual earnings growth for 7-8 years - still below Nu's 40%+ trailing pace the last two quarters, though the gap narrowed slightly this quarter as the multiple expanded faster than earnings did.

Nu's stock continued the two-year climb described in nu/2025-06: from roughly $8.20 (October 2023) it rose to $16.01 by the close of this quarter (September 30, 2025) - more than double over two years, but with the same intra-period volatility (a dip to $10.24 in March 2025) that makes a single point-to-point read less informative than the underlying earnings trend. With approximately 4.91 billion diluted weighted-average shares outstanding:

Market cap → enterprise value	Q3 2025
Share price (period-end)	$16.01
Shares outstanding	~4.91B
Market capitalization	~$78.5B
Total liabilities	n/a (net cash basis, minimal net debt - see nu/2025-06)
Less: cash and equivalents	n/a
Enterprise value	~$78.5B (approx., minimal net debt)

TTM net income (Q4'24 through Q3'25, summing parent-attributable figures: $552.6M + $557.2M + $637.0M + $782.7M) is approximately $2,529.5 million.

Peer-multiple sanity check	Q2 2025	Q3 2025	Change
Enterprise value / market cap	~$67.2B	~$78.5B	⚠️ up
P/E (trailing)	~29.2x	~31.0x	⚠️ up slightly

31x trailing earnings remains a large premium to Brazilian incumbents Itaú Unibanco and Banco Bradesco (typically 8-10x trailing), consistent with the growth-rate gap between Nu and those banks, and roughly in line with where Nu itself traded the prior quarter.

Reverse DCF: at $78.5 billion and ~$2.5 billion TTM net income, using an 11-12% discount rate appropriate for Nu's Brazil/Mexico/Colombia currency and regulatory exposure, the market is pricing in roughly 20-23% sustained annual earnings growth for 7-8 years before a terminal multiple of 12-15x. Nu's trailing net income growth has been running 40%+ YoY the last two quarters - still comfortably ahead of what's priced in, which keeps the bull case intact, though the gap between actual and priced-in growth narrowed slightly this quarter as the multiple expanded faster than earnings did.

The reverse-DCF and peer-multiple lenses are the more honest read here.

---