Full Report

Industry — Understanding the Playing Field

1. Industry in One Page

Latin American retail banking is a deposit-funded consumer-credit business wrapped in a payments and fee layer. Five incumbents in Brazil, Mexico and Colombia have historically held 68%–80% of all loans and deposits in each market and earned high spreads because branch-heavy distribution kept new entrants out and tens of millions of adults underserved. Smartphone penetration, instant-payment rails (Pix, CoDi, Bre-B) and Open Finance reset the entry cost: a mobile-first balance sheet can underwrite, fund and service customers at a fraction of the legacy cost. The retail financial-services revenue pool across Brazil, Mexico and Colombia was $227.9 billion in 2025 (interest income + fees, net of funding), growing 7% reported and 11% on an FX-neutral basis. This is not a payments app: interest income on credit-card and loan balances funded by sticky retail deposits is where 80%+ of the money is made. Payments and fees buy engagement and data; spread on the lending book is the profit engine.

Takeaway: profits stack across the value chain, but credit underwriting on top of cheap deposits is where banks earn their spread; everything else either feeds that engine with data or monetizes it with fees.

2. How This Industry Makes Money

Retail banks earn a net interest margin — the gap between what they charge on credit-card balances and loans (often 30%–400% APR in Brazil) and what they pay for funding (deposits, wholesale debt, capital). Out of that gap come expected credit losses (provisions for loans that won't be repaid), transactional costs (interchange paid to networks, fraud), and operating expenses (branches, staff, technology, marketing). What's left is pre-tax income. A second leg — fee and commission income — is generated by interchange on debit/prepaid spend, brokerage and asset management, insurance distribution, and increasingly subscription-style products. For Nu specifically, interest income was 85% of 2025 revenue and fees 15%; that ratio is broadly representative of digitally-led retail banks that lend.

Two structural cost lines dominate any retail bank: funding cost (about 29% of Nu's 2025 revenue) and credit losses (about 27%). Combined, they consume roughly half of every revenue dollar before a single branch or engineer is paid. This is why "cost of funding" and "cost of risk" are the two metrics that distinguish a great bank from an average one — saving a few hundred basis points on either flows almost entirely to pre-tax income.

Bargaining power sits in three places:

• Depositors: in markets with high policy rates (Brazil Selic averaged 14.3% in 2025), depositors can demand close to the interbank rate. A bank that brands itself as the customer's "primary relationship" (rather than a one-time deposit) earns funding below 100% of the interbank rate. Nu cites 81% of CDI in December 2025.
• Card networks and merchants: interchange is set by Mastercard/Visa and paid by merchants. Banks distribute the cards and receive the bulk of interchange. Caps or reforms here directly hit fee income.
• Regulators: they set capital requirements, interchange rules, and — in Brazil since 2023 — caps on revolving credit-card interest. Each can compress a layer of the margin stack overnight.

Capital intensity is high in the regulatory sense (banks must hold equity against risk-weighted assets) but low in the physical sense for a digital model — branches and back-office headcount are the legacy fixed cost the new entrants are demolishing.

3. Demand, Supply, and the Cycle

Retail-banking demand is shaped by three structural forces and one cyclical one. Structural: smartphone penetration (GSMA forecasts 93% in Latin America by 2030), formal-credit expansion among populations that have been historically unbanked, and migration of payment volume from cash to instant electronic rails. Cyclical: the policy-rate cycle. When central-bank rates rise, banks' funding costs rise immediately while loan repricing lags, compressing spread; rising rates also pull deposits out of low-yield savings into higher-yield products, raising the cost of funding. When rates fall, the dynamic reverses and credit demand picks up.

Sources: 20-F citations of World Bank, BIS, IMF, BCB, Banxico, ABECS data; rates and unbanked figures vary by source date.

The under-penetration is striking: Mexico's credit-card penetration of 11.3% sits four times below Brazil's, and household debt at 17% of GDP is roughly a quarter of the developed-market norm. That gap is the demand backdrop for a generation of credit growth. But it isn't free money — high policy rates raise both opportunity cost for depositors and credit-loss potential for the lender.

Where the cycle hits first in a retail bank is interest expense on deposits (immediate, on every dollar of funding) and expected credit loss provisions (immediate, forward-looking under IFRS 9). Loan-book yield repricing comes later. That is why digital-bank quarterly results in Brazil swing on the trajectory of the Selic and on shifts in NPL formation in unsecured consumer credit; both data points are released monthly by the central bank.

4. Competitive Structure

Brazilian, Mexican and Colombian retail banking is one of the most concentrated banking markets in the world. The Herfindahl–Hirschman Index (a 0–10,000 measure of concentration where higher = more concentrated) sits at 889 in Brazil, 993 in Mexico and 1,262 in Colombia, versus 328 in the US, 373 in Germany and 488 in France. The top five incumbents in each market hold between 68% and 80% of all loans and deposits. That concentration is exactly why disruption has been possible — it created persistent excess profit and customer dissatisfaction that a low-cost digital entrant could attack.

The structure is oligopolistic on the incumbent side and fragmented on the challenger side. Disruption has come from three distinct types of entrants — pure digital banks, payments-led fintechs that built deposit and credit franchises on top of merchant relationships, and e-commerce platforms that bolted financial services onto their user base. Each subtype has a different cost structure and a different right to win.

Two facts about the structure deserve emphasis. First, digital banks' share of total outstanding loans to Brazilian individuals went from below 1% in 2018 to above 8% by September 2025 (and above 14% if mortgages and earmarked rural credit are excluded). That is a structural shift, not a cycle. Second, the incumbent block is not monolithic — Itaú reported a 2025 net interest margin of 3.6%, well below its 2024 4.4%, and incumbent ROEs have come under pressure as digital competitors have peeled off the highest-spread, most-profitable consumer segments first. Concentration is high, but it is also actively contested.

5. Regulation, Technology, and Rules of the Game

Latin American regulators are unusually proactive in trying to engineer competition into a concentrated industry. The clearest cases — Brazil's Pix instant-payment system (October 2020) and Open Finance (rolled out from 2021) — were explicit central-bank decisions to lower switching costs and force interoperability. Pix reached 179.7 million unique users and $6.3 trillion of transaction volume in 2025, a 29% reported increase over 2024 (34% FX-neutral) and the dominant payment rail across consumer and SME use cases. Mexico has copied the playbook with CoDi (2019) and DiMo (2023); Colombia launched Bre-B as its low-value instant-payments system. The economic effect is consistent — cash and cheque float migrate to electronic balances held in mobile-first apps, and the incumbents' interchange and float advantages erode.

Technology is the second rule-changer. The 2010s shift from on-premise core-banking systems to cloud-native ones eliminated the cost moat that incumbent IT systems used to provide. Nu's NuCore platform — proprietary, cloud-based, built in Clojure — illustrates what the marginal entrant can now do. The 2024 acquisition of Hyperplane added an AI layer for personalization that incumbents will struggle to replicate without rebuilding their data foundations. AI- and ML-driven underwriting compounds: more customers generate more data, which sharpens credit decisions, which lowers cost of risk, which funds lower customer-facing prices, which acquires more customers.

6. The Metrics Professionals Watch

Retail banking has its own vocabulary. The metrics below explain value creation in this industry better than headline EPS — beginner analysts should ground a model on these before tracing them up to net income.

The most useful skeptical check on any LatAm digital-bank narrative is to triangulate three of these: (1) cost of funding as % of policy rate, (2) NPL formation in the unsecured book, and (3) ARPAC against cost-to-serve. A bank that is widening its ARPAC–cost spread while keeping cost of funding well below 100% of the interbank rate and NPLs stable is creating real value. A bank that is growing customers fast but with deteriorating funding cost or unsecured-loan NPLs is buying revenue.

7. Where Nu Holdings Ltd. Fits

Nu is a digital-native challenger that has reached scale — large enough by customers to be the #1 private financial institution in Brazil (113 million customers, 62% of adults), the leading new-credit-card issuer in Mexico, and a meaningful entrant in Colombia. It sits on the digital-bank rung of the industry ladder defined above, but with a balance sheet (deposits $41.9B, credit-card receivables $18.3B, loans $9.4B at FY2025) that increasingly looks like an incumbent's by composition, while operating costs look like a tech company's. The US OCC conditional national-bank charter granted on January 29, 2026 marks an intended step outside Latin America, into the US digital-banking market.

8. What to Watch First

A short checklist of observable signals that will tell a reader whether the industry backdrop is improving or deteriorating for Nu, in roughly the order they move:

1. Brazil Selic and cost of funding as % of CDI. Monthly BCB releases. A widening gap between Nu's funding cost and CDI (currently 81%) compresses NIM directly. A narrowing gap is the single best operational signal.
2. NPL 15–90 and NPL 90+ in unsecured Brazilian consumer credit. BCB monthly statistics plus Nu's own quarterly disclosure (3.9% / 6.6% as of December 2025). A meaningful uptick in the 15–90 bucket is the leading indicator of credit-loss provisions one to two quarters ahead.
3. Pix transaction volume and Open Finance call volumes. Both expand the addressable deposit pool and lower switching costs. Quarterly BCB reports.
4. Brazilian and Mexican incumbent bank ROEs and NIMs. Itaú, Bradesco, BBVA México disclosures every quarter. Persistent ROE/NIM compression at incumbents signals share-shift continuing.
5. Digital banks' share of total Brazilian consumer loans. Reported by BCB; currently around 8%, with the trajectory mattering more than the level.
6. Brazilian regulatory developments on revolving credit-card cap, RWA methodology, and Mexican banking license issuance for Nu. Each can reshape spread or balance-sheet capacity in a single notice.
7. Updates on Nu's OCC pre-opening conditions and FDIC insurance application for Nubank, N.A. The US charter is the only catalyst with multi-year TAM implications outside Latin America.

If those seven move favorably together, the industry backdrop supports Nu and the rest of the report can be read with a tailwind. If three or more move against the company in the same quarter — particularly a Selic spike combined with rising NPLs — the industry is working against the franchise, not with it.

Know the Business

Nu is a spread lender wearing a tech wrapper: 85% of revenue is interest income on credit-card and personal-loan balances, funded by sticky retail deposits at 81% of the Brazilian interbank rate, distributed through one app at a monthly cost-to-serve of about $0.80 per active customer. The moat is unit cost, not pricing — and the moat compounds because every additional customer makes the underwriting models smarter without adding meaningful overhead. The market is correctly pricing a high-quality compounder (P/E ~20, P/B ~5, FY2025 ROE 30%); what it may be underestimating is the gap between blended monthly ARPAC of $15 and mature-cohort ARPAC of $27, and what it may be overestimating is the durability of a 26.7%-of-revenue cost-of-risk line in a Brazilian unsecured-credit cycle the company has not yet been tested through at this scale.

How This Business Actually Works

One sentence: Nu rents money from depositors at sub-policy-rate pricing, lends it to consumers at credit-card and personal-loan rates, eats a 27%-of-revenue credit-loss line, and keeps the difference — because operating costs are 19.9% of revenue, less than half of what an incumbent's branch network costs.

The chart on the right is the most important picture in this report: funding cost plus credit losses consume 56 cents of every revenue dollar before a single employee is paid. Marketing at 1.9% of revenue and customer support at 4.1% are rounding errors next to a 1 pp move in cost of funding or NPL formation. This is why "tech multiple" framing is misleading — earnings variance is driven by two bank lines, not software cost curves.

Bargaining power is asymmetric and sits in three places. Customers have very little: switching is technically easier than ever with Open Finance, but Nu's NPS above 85 and primary-banking-relationship penetration over 60% mean the friction is psychological, not structural. Depositors have meaningful power in a high-Selic regime — Nu pays 81% of CDI today, and that ratio is the single best operational signal for whether the franchise is intensifying or weakening. Regulators hold the most leverage: a revolving-credit interest cap (already in force since 2023), an RWA methodology revision, or an interchange reset can compress a layer of the margin stack overnight.

The Playing Field

Three things this peer set reveals. First, Nu is alone in its quadrant: only MELI matches the ROE-and-growth profile, but MELI is an e-commerce platform earning bank-like returns from a non-bank base, priced at 11.7x book versus Nu's 5.2x. Among pure financial peers, the incumbents (ITUB, BBD) trade at 2x book or less because their ROE is mid-to-high teens with no growth, and digital challengers (INTR, PAGS) trade below 1.5x book because they have not yet shown they can earn an incumbent-class ROE. Nu has done both — 30% ROE with 27% revenue growth — and is the only LatAm bank for which a 5x book multiple is rationally defensible. Second, what "good" looks like in this industry is converging Itaú-level ROE (around 20%) with digital-bank cost structure; Nu has already overshot the ROE benchmark by ~10 points and has further unit-cost room to run. Third, the bear case sits in the peer table too: Inter and PagBank are also digital, also Brazilian, also Open Finance-enabled — and their ROEs are 14% and 14.5%. The premium ROE Nu earns is not a structural property of being digital; it is the property of being the digital franchise that achieved scale first.

Is This Business Cyclical?

Moderately to highly cyclical, with the cycle hitting through cost of risk and cost of funding before it touches revenue. Nu has never reported through a deep Brazilian unsecured-credit downturn at current scale — it became profitable only in FY2023, and the 2023 NPL spike (Brazil's last real consumer credit stress) was managed without distress, but the loan book has grown roughly 4x since.

The provision-growth chart tells the most underappreciated story in the Nu file: the 2022 232% jump in provisions was the company's stress test, and it survived without distress — but the trajectory of 33% provision growth in FY2025 against 27% revenue growth means cost of risk is still rising slightly faster than the topline. That is normal in a growing unsecured book, but it is the line to monitor: a flip where provisions outpace revenue by 10+ pp in a single year would be the signal that the cycle has turned. Nu's reported leading indicator (15-90 NPL at 3.9% in December 2025, down from 4.1% at Q3) is the cleanest real-time gauge.

The Metrics That Actually Matter

Forget headline EPS. Five numbers will tell you whether Nu is creating or destroying value, in roughly this order of importance:

The two metrics worth more attention than they get are ARPAC and cost-to-serve, together, because their spread is the unit economic. ARPAC has tripled from $4.50 to $15 since 2019; cost-to-serve is essentially flat at $0.80; mature cohorts are already at $27. If a mid-cohort customer worth ~$15 in revenue today and ~$2 in net contribution can plausibly become a $27-revenue/$2-cost customer over a decade, the company's intrinsic value compounds well above what consensus is modeling — because the marginal customer needs no capex or branch.

What Is This Business Worth?

The right lens is P/B against sustainable ROE, with explicit modeling of the ARPAC compounding curve and cost-of-risk normalization through a cycle. This is one economic engine, not a sum-of-parts: Brazil, Mexico and Colombia are the same playbook at different stages of customer-base maturation, and the consolidated numbers do not hide materially different businesses. SOTP would force-fit segment math the company does not disclose at the necessary granularity.

The peer P/B-vs-ROE chart frames the valuation question precisely. If Nu's sustainable through-cycle ROE is 30%, the 5.2x book multiple is rational versus the bank-peer line that prices Itaú at 2.2x for 21% ROE. If sustainable ROE is closer to 20% — i.e., cost of risk normalizes higher in a cycle and ARPAC compounding slows — the multiple should compress toward 3x book, implying meaningful downside even before any growth disappointment. The asymmetry: ARPAC has a clear path higher (mature cohorts at $27 vs blended $15) and customer growth has multi-year runway; both push the numerator up. The risk: cost of risk is more sensitive to Brazilian unemployment than to anything in management's control.

What would change the thesis materially: (1) cost of funding rising above 90% of CDI for two consecutive quarters; (2) 15-90 NPL rising above 5% with no seasonality explanation; (3) Mexico/US capital absorption pushing equity growth below earnings retention; (4) revolving-rate or RWA regulatory move that compresses NIM by 100+ bps. Absent these, the franchise is doing what a high-quality compounder does and the multiple is defensible.

What I'd Tell a Young Analyst

Stop thinking of this as a tech stock and stop thinking of it as a bank. It is a digital balance sheet — a bank by economics, a tech company by cost structure. The dollar of revenue you should model is not "ARPU times users"; it is (spread on credit balances + fees per active customer) minus (cost of funding + cost of risk + $0.80 cost-to-serve). Everything else is decoration.

Three watch items, in priority order:

1. The ARPAC-versus-cost-to-serve gap. If ARPAC compounds at 15%+ per year and cost-to-serve stays under $1, every other concern is secondary. If cost-to-serve creeps up — because Mexico support takes more headcount, or US compliance adds layers — flag it immediately; the unit economic is the franchise.

2. 15-90 NPL, monthly. This is the leading indicator that prints before earnings. Brazil's BCB publishes consumer-credit NPLs monthly; Nu's own disclosure is quarterly. Watch the BCB series in the months between Nu prints — a 30 bps move in 15-90 NPL on the unsecured book is the kind of early signal that moves the stock when the next quarter lands.

3. Cost of funding as % of CDI. Today 81%, the company's chosen ceiling. If a quarterly print comes through at 85%+ without a corresponding 100 bps spike in Selic, that is the deposit franchise weakening — possibly because Inter, Mercado Pago or BTG Digital is bidding deposits up. This is the metric where the moat shows up first, and where the first crack would appear.

What the market may be missing: mature customer cohorts already generate $27/month versus a $15 blended average — that is not aspirational, it is a fact in 2025. If the rest of the base ages into that curve, FY2030 ARPAC could be 60%+ above today on essentially the same cost-to-serve. What the market may be overestimating: that Mexico replicates Brazil at Brazil's profitability. Mexico has Mercado Pago co-leading the market, a different policy-rate dynamic, and Sofipo-licensed competitors with deposit rates that already eat into Nu's cost-of-funding edge. The Brazil playbook will not run with Brazil's economics.

What would actually change the thesis is not a quarterly miss on ARPAC or a single bad NPL print — it is structural: the OCC charter consuming more capital than the LatAm franchise generates, a Brazilian recession driving 90+ NPL past 9%, or a revolving-rate cap that compresses card NIM by 200+ bps. Those are the events to underwrite against.

Quality Score — Nu Holdings Ltd. (NU)

The quality profile reads as BA — the business clears the durability bar on a 30% ROE / 41% net margin / 1.22x cash-conversion footprint, and the same founder-led team that built the franchise has reinvested every dollar at above-incumbent returns without diluting holders. The binding offsets are valuation (priced for the quality even after the May 14–15 drawdown) and geography (90% Brazil), neither of which compresses the underlying compounder math.

Where the evidence agrees and where it doesn”t

The strongest cross-specialist consensus sits on Business Quality (B) and Allocation (A). Numbers, Warren, Moat, and Forensic all converge on a 30% ROE / 41% net-margin / 1.22x cash-conversion compounder with a narrow but real moat in unit cost (14k customers/FTE) and deposit funding (81% of CDI). Historian closes the loop on A: the current leadership did not inherit this business — they built it across 2013–2021 — and they have retained every dollar of $3.5B FCF at 30%+ ROE with only ~1% annual dilution. That is a clean BA setup, and Stan”s "Lean Long, Wait For Confirmation" verdict is consistent with it.

The sharpest disagreement is concentrated on Valuation (V) and the Specialness (S) subclaim around cost-of-funding. Bull target $18 vs Bear target $7.50 is a >2x spread on the same disclosed numbers — Variant flags 4 active disagreements at 62 strength and 78 consensus-clarity. The single most important S-dimension challenge: Inter & Co”s FY25 20-F discloses funding at 65.3% of CDI versus Nu”s 81%, a 160 bps disclosure-level gap that suggests the deposit-cost moat is narrower than consensus models. We held S at 4 (distinctive but with available substitution) rather than 5 because that gap is real on disclosed data.

The weakest dimensions to monitor are V and G. V moves up to 4 if the Q2 FY2026 print (week of Aug 13–14) clears provisions ≤ $1.6B with ECL coverage held at 15.4% without strain — that would re-couple the multiple to the compounder engine. V moves down to 2 if Q2 forces a coverage uplift through the P&L while Mexico provisions stay elevated. G is structurally capped at 3 until Mexico crosses 15–20% of revenue or the US bank capitalizes within the OCC”s Jan 29, 2027 deadline without absorbing equity. The thinnest evidence is on management”s execution credibility outside Brazil (7-of-11 promises kept, Mexico provisioning still maturing), which is why A is rated medium-confidence rather than high.

Competition — Who Can Hurt Nu, Who Nu Can Beat

Competitive Bottom Line

Nu has a real, measurable moat — but it is a unit-cost moat, not a product moat, and the company most likely to test it is Mercado Pago (MELI), not the Brazilian incumbents Nu is currently taking share from. Against incumbents (Itaú, Bradesco), Nu wins on cost-to-serve, brand, and growth and is converting share at a pace BCB data confirms (digital banks' share of Brazilian individual loans went from under 1% in 2018 to above 8% by September 2025; Itaú's individual-loans share sits at 10.7%). Against pure-play digital peers (Inter, PagBank), Nu is the only one earning a 30% ROE — Inter is at 14.4% and PagBank at 14.5%, which is the point that should be most uncomfortable for Nu shorts. The competitor that breaks the easy narrative is Mercado Pago: it earns a 36% ROE off an e-commerce ecosystem Nu cannot reach, is the #1 fintech by MAUs in Argentina/Chile/Mexico, and is the only meaningful entity that out-grows and out-monetizes Nu in Mexico, the geography that has to work for the next leg of the equity story. Brazil is Nu's to defend; Mexico is contested.

The Right Peer Set

The five competitors below were chosen to span Nu's three competitive lanes: (1) the Brazilian incumbents whose deposit and credit-card share Nu is taking (ITUB, BBD); (2) the digital-bank pure-play that most closely shares Nu's operating model (INTR); and (3) the LatAm fintech/payments platforms with the most consumer-credit and digital-account overlap (MELI, PAGS). StoneCo (STNE) appears in the valuation table as a secondary payments comparable but is not in the threat map because its SMB-acquirer and Linx software stack overlaps with Nu only at the margin.

Valuation Snapshot (as of 2026-05-15)

The picture this peer table draws is the heart of the moat question: Nu is the only listed LatAm bank in the high-ROE / high-growth / high-P/B quadrant, and it is also the only one that is taking share from the incumbents at the rate the BCB data reveals. ITUB and BBD trade at 0.5x–2.1x book because their ROE has plateaued in the mid-teens to low-twenties with no growth. INTR and PAGS — the obvious "Nu, but smaller" trades — earn 14% ROEs and are priced accordingly. MELI is the only structural exception: 36% ROE off an e-commerce flywheel, priced at 15x book. The competition tab's central question is whether Nu's economics are durable enough to remain in that lonely quadrant.

Where The Company Wins

Four advantages are visible in the data, not just in marketing decks. Each is supported by either Nu's own FY2025 20-F, the peer 20-Fs filed under the same SEC framework, or independently reported BCB / CNBV statistics.

1. Unit cost — ~14,000 customers per employee vs. ~1,200 at incumbents

The single most important fact in the peer file: Nu serves ~14,000 customers per employee versus an incumbent average of ~1,200. That ratio drives Nu's 19.9% efficiency-ratio target and roughly $0.80/month cost-to-serve, both at least 60% below Itaú and Bradesco's branch-loaded equivalents. None of the digital peers (Inter, PagBank) match it because they have not reached Nu's customer scale — Inter at 25M active vs. Nu at ~83M monthly active. The unit-cost advantage is the moat, and it compounds because the cost line is fixed in technology, not variable in headcount.

Sources: Nu FY25 20-F + Industry tab; INTR FY25 20-F (R$24B opex / 43M clients / ~7k FTEs ≈ 6,000); ITUB, BBD efficiency ratios from FY25 20-F MD&A. Employee ratios are approximate; cost-to-serve is Nu's own disclosure.

2. Brand-led customer scale that outruns the incumbents

Customer growth at Nu in 2025 — 17 million net adds, to 131 million customers, growing 65% of Brazilian adults — exceeded the combined net adds of the top-5 Brazilian incumbents per BCB monthly statistics. Nu's NPS sits above 85 in Brazil, well above Itaú's Uniclass affluent-tier 76 (and Uniclass is Itaú's best segment). Customer base is the input to every other line in the model: deposits, ARPAC, cross-sell, and underwriting data quality. None of the peers have closed this gap — Inter has reached 43M clients but is growing slower in absolute terms.

Itaú Uniclass NPS 76 is the bank's affluent-segment number from its FY25 20-F — the closest disclosed equivalent. Mercado Pago does not separately disclose Brazilian customer counts in 10-K business section.

3. Brazil credit-card share — Nu is now the #2 issuer by purchase volume

Per independently reported BCB data, Nu holds approximately 24.3% of Brazilian credit-card purchase volume (up roughly 50 bps in Q4 2025), versus Bradesco's 12.5% and Itaú at the high end of the incumbent range. This is the share that translates most directly into interest income because the credit-card book funds the highest-yield revolving and installment balances on the asset side. Both Itaú and Bradesco continue to lose interchange/fee mix to Nu — and the regulators' 2023 cap on revolving credit-card interest hit incumbents harder because they have higher-revolving mix than Nu's installment-led portfolio.

4. Mexico — already the #3 credit-card issuer (CNBV data)

Nu Mexico's credit-card customer base reached 6.6 million by June 2025, up 52% year-over-year, making Nu the #3 credit-card issuer in Mexico per CNBV — behind only BBVA México (10.7M cards) and Banamex (9.2M cards). Translated to cards-in-force share, Nu sits at roughly 17–18% of the 37.1M total Mexican credit-card market, an order of magnitude above where it was two years earlier. Mercado Pago is competing for the same wallet, but Nu has the issuance volume and a banking-license pathway (Nubank S.A. in formation). 22% of Nu's Mexican customer base entered the formal financial system for the first time through Nu — that is a sticky relationship.

Where Competitors Are Better

The honest answer is that several peers have at least one dimension where they are visibly stronger than Nu — and ignoring those gaps overstates the moat.

1. Inter has a structurally lower cost of funding (65.3% of CDI vs. Nu's 81%)

Inter & Co's FY2025 20-F discloses cost of funding at 9.3%, or 65.3% of the average CDI in 2025. Nu pays roughly 81% of CDI (December 2025 disclosure). On a like-for-like basis, Inter funds itself ~160 bps cheaper than Nu. Inter attributes this to a higher concentration of deposits in total funding — and it is the single uncomfortable fact in the peer set, because deposit-franchise quality is exactly what Nu's bull thesis assumes is best-in-class. Inter is much smaller (R$55B funding vs. Nu's $41.9B deposits in USD, ~R$240B), so the question is whether scale forces Nu to pay up for deposits as the book grows. The 81% figure has been creeping up from the high 70s — that is the metric to watch.

2. Itaú still has the corporate, mortgage, and payroll relationships that Nu cannot reach

Itaú Unibanco is the bank that has actually managed the digital transition best among incumbents — 21% FY25 ROE, ~71M retail customers, and a wholesale/SME franchise Nu has barely entered. Itaú's R$1.9 trillion balance sheet funds segments where Nu has no presence: corporate lending, large-ticket mortgages, FX/trade finance, and asset management for high-net-worth. Itaú Uniclass (affluent segment) reports a 76 NPS — a number that should give pause to anyone modeling Nu's eventual move upmarket. The high-ARPAC affluent customer is contested ground that Itaú already dominates.

3. Bradesco controls the insurance-distribution rent (22.8% market share)

Bradesco's 22.8% share of the Brazilian insurance market (per SUSEP/ANS as of September 2025), through Grupo Bradesco Seguros, is a capital-light, high-margin annuity that Nu's nascent NuLife product is years from replicating. Insurance distribution is exactly the kind of low-incremental-cost fee revenue Nu's bull case requires to drive ARPAC from $15 to mature-cohort $27 — and it is the line where Bradesco's branch network has the most durable advantage. R$118.5 billion of 2025 insurance premiums + pension + capitalization is a moat in its own right.

4. MercadoLibre's ecosystem economics — 15x P/B is not just exuberance

MELI earned a 36% ROE on FY25 revenue growth of 39% — operating an e-commerce marketplace that funds a payments rail that funds a credit book Nu cannot replicate from a banking license alone. Mercado Pago is the #1 fintech by MAU in Argentina, Chile and Mexico, and #2 in Brazil. In Mexico — Nu's most important growth market — Mercado Pago has roughly comparable digital-wallet penetration, a deposit and credit franchise growing fast, and the e-commerce traffic Nu must acquire through marketing. MELI also trades at 15x book versus Nu's 5.2x, indicating the market gives more credence to the e-commerce flywheel than to a pure-bank model. The single peer Nu cannot easily catch is MELI.

5. PagSeguro/PagBank owns the merchant-acquiring lane Nu has not entered

PAGS has 6.3 million active merchants and serves 34 million clients across acquiring + PagBank. Nu launched a Working Capital SMB product in Brazil in 2024 and a Charging Assistant in 2025 — both are early. Anywhere Nu's SMB cross-sell roadmap depends on payment-acquiring reach, PagBank and StoneCo have a head start. The economics are also worse on the acquiring side (commoditizing rates), which is why these stocks trade at sub-1x book — but for SMB-customer ownership, the head start is real.

The heatmap is deliberately uncomfortable for the Nu bull case in a few rows: Mercado Pago beats Nu on Mexico position and ARPAC monetization velocity, Bradesco beats Nu on insurance distribution, Itaú beats Nu on corporate/affluent banking, PagSeguro beats Nu on merchant acquiring, and Inter beats Nu on cost of funding. Nu's win is the columns no other peer dominates — unit cost, Brazil consumer-credit share, NPS, and the combination of growth and ROE.

Threat Map

Five threats are concrete enough to monitor. Severity is anchored on (a) the probability the threat materializes inside 24 months and (b) how much earnings power Nu would lose if it did. "Pricing-pressure" threats rank higher than "narrative" threats because the deposit-franchise math is unforgiving.

Moat Watchpoints

Five measurable signals will tell an investor — quarter-by-quarter — whether the moat is widening, holding, or narrowing. Each is observable in either Nu's own disclosure, BCB monthly statistics, CNBV Mexico statistics, or peer 20-F equivalents.

Current Setup & Catalysts — Nu Holdings Ltd. (NU)

1. Current Setup in One Page

The stock is trading at $12.07 — down 36% peak-to-trough from the February 3, 2026 high of $18.83 — and the market is mostly watching one number: whether the Q1 2026 expected-credit-loss spike (+31% QoQ to $1.72B) reverses, holds, or accelerates in the Q2 2026 print expected the week of August 13–14, 2026. The setup is mixed-to-bearish: revenue beat at a record $5.0B with net income still +41% YoY, but a 1¢ EPS miss against $0.20 consensus, risk-adjusted NIM contracting 100 bps to 9.5%, and a 89-bp jump in 15–90 day NPL to 5.0% triggered a 7% one-day drop on May 15 and a death-cross confirmation on the 200-day moving average back on April 15. Two heavyweight setup items were absorbed earlier in the year — the OCC conditional charter for Nubank, N.A. on January 29 and Brazil's first Copom cut in five meetings (Selic 15% → 14.75% on March 18) — but the market has retreated to a single question that only Q2 can resolve. The forward calendar is medium quality: one decision-grade hard date (Q2 print in ~90 days), two policy windows that can move the tape (Copom in late-June/late-July; Brazil first-round presidential election October 4, 2026), and an OCC-mandated capitalization deadline for Nubank, N.A. in January 2027.

Recent Setup Rating: Mixed

2. What Changed in the Last 3–6 Months

The narrative arc, in one paragraph. Six months ago the market was paying ~28x P/E and 7.2x P/B for a 30%+ ROE compounder, with the bullish overlay anchored on three pillars: Berkshire's silent endorsement, Mexico approaching break-even, and a US-charter optionality bet that had not yet been priced. Since then, two of those pillars wobbled — Berkshire is gone per the 13F record, and the US charter has been re-read by part of the sell-side as a capital absorption rather than a free option. The third pillar held (Mexico hit break-even Q1 2026 with 15M customers, #3 by user count). What replaced the old debate is a single binary: did the Q1 ECL jump represent disciplined provisioning into a maturing book, or did the 56% one-year loan growth at flat 15.4% ECL coverage turn out to be the late-cycle leverage point the bear case warned about? That question is unresolved going into Q2.

3. What the Market Is Watching Now

The live debate, distilled: is Nu a 30% ROE compounder pausing on seasonality, or a 20-22% ROE Brazilian credit lender that the market over-priced at 7x book? Every catalyst over the next six months either reinforces one interpretation or settles it. The market is essentially trading a single instrument — Q2 provisions and the 90+ NPL print — even though there are at least five separate event lines that can move the stock.

4. Ranked Catalyst Timeline

5. Impact Matrix

6. Next 90 Days

The 90-day window has one decision-grade event (Q2 print) and two policy-window events (Copom × 2) that frame it. Sell-side preview season (late July) is the soft window where the consensus EPS line will likely move — that is the leading indicator the tape historically respects. Anything beyond 90 days (the October election, the Q3 print, the OCC capitalization deadline) is real and dated but is not what an underwriter sets a clock on today.

7. What Would Change the View

The view changes on three observable signals, in order of priority. First, the Q2 2026 provisions print and 90+ NPL ratio — a sub-$1.6B provision figure with stable 90+ NPL below 6.6% closes the cost-of-risk debate that the May 15 reset opened and re-couples the stock to its 30% ROE compounding engine; a sustained $1.7B-plus print with 90+ NPL above 7% on flat 15.4% coverage validates the bear thesis that the ECL ratio is mechanical and forces a 100-200 bps cost-of-risk reset that takes ROE to incumbent levels. Second, the Brazilian rate path through the October election — if Copom delivers 75-100 bps of cumulative cuts through Q3 without an election-driven pause, the funding-cost and loan-demand tailwinds feed directly into the cohort ARPAC engine the bull case rests on; if election politics force a Copom reversal or a sustained BRL weakness, the entire LatAm banking cohort re-rates and Q2 fundamentals will not save the stock. Third, the US bank capitalization disclosure — quarter-by-quarter capital deployment that stays inside the FCF generation envelope (and inside the ≤100 bps FY26 efficiency-ratio guide) validates management's framing of the charter as a bounded optionality bet; any equity issuance, dividend/buyback freeze, or material CET1 compression linked to Nubank, N.A. converts the bull thesis on capital allocation into the bear thesis on capital absorption. These three together are the event path the next six months will trade on; everything else — Mexico disclosure, ownership rotation, Open Finance Phase 4 — is information that calibrates the call but does not force the debate to update.

Bull and Bear

Verdict: Lean Long, Wait For Confirmation — the May 14–15 reset to 5.2x book on a 30% ROE franchise creates a credible entry, but the Q1 2026 provisioning crack and the Bear's "mechanical ECL coverage" claim are unresolved on the data we have today. Bull carries the valuation and quality argument; Bear carries the near-term cost-of-risk argument and one structural objection (Inter's 160 bps cheaper funding) that Bull does not refute. The decisive tension is whether the flat 15.4% expected-credit-loss coverage held through 56% loan-book growth is analytical underwriting (Bull) or a denominator artifact (Bear) — both sides cite the same disclosed ratio and reach opposite conclusions. The Q2 FY2026 print in mid-August is the falsifying event for both theses; we would not chase the stock ahead of it, and we would not short a 30% ROE franchise into it either.

Bull Case

Bull target. $18.00 per share via 6.0x forward TBV on FY25 TBV/share of $2.09 compounded at ~33% (the FY23–FY25 CAGR, conservative against the 49% FY25 print) to ~$2.78–$3.00 at end-FY26. Cross-checks at ~22x forward P/E on FY26 EPS of ~$0.82, in line with peer growth-banks and well below the 28.6x NU printed at FY25 close. Timeline: 12–18 months. Primary catalyst: Q2 FY2026 print (mid-August 2026) showing 90+ NPL at or below 6.6% with ECL coverage maintained at ≥15.4%. Disconfirming signal: Brazil 90+ NPL above 8% in any quarter through Q4 FY2026 with ECL coverage held flat — that combination would force a 100–200 bps cost-of-risk reset and ROE below 22%.

Bear Case

Bear target. $7.50 per share via P/B compression to 3.2x on YE25 book value of ~$2.30/share, justified by ROE normalization to 22–25% on through-cycle cost of risk and Mexico/US capital absorption, anchored to Itaú's 2.1x book / 21% ROE peer benchmark plus a residual growth premium and a governance/disclosure haircut (74% voting control by Vélez on 18.6% economics, Cayman foreign-private-issuer status, new "Managerial P&L" framework under KPMG limited assurance). Timeline: 12–18 months. Primary trigger: Q2 FY2026 provisions sustained at or above $1.6B with no coverage-ratio uplift above 15.4%, or BCB monthly 15-90 NPL above 4.5% before the print; October 2026 Brazilian presidential election as parallel volatility. Cover signal: Q2 FY2026 provisions print below $1.4B and 15-90 NPL declines below 3.5% — that combination invalidates Mexico-seasoning-becomes-permanent and reasserts the cohort engine.

The Real Debate

Verdict

Lean Long, Wait For Confirmation. Bull carries more weight on the franchise and the entry: a disclosed 30.3% ROE on 6.6x assets-to-equity with $41.9B of deposits funding a $27.7B loan book and $3.49B of FCF is genuinely rare in LatAm banking, and a multiple reset from 7.2x to 5.2x book on a 1¢ EPS miss is not a thesis-breaker, it is a price improvement. The decisive tension is the 15.4% ECL coverage line — Bear is right that an exactly-flat ratio across two years and 56% loan growth deserves scrutiny, and right that the Q1 2026 provisioning jump and Inter's 160 bps cheaper funding are real disclosure-level objections the Bull case does not refute on the available evidence. Bear could still win the trade if Q2 FY2026 forces a coverage uplift through the P&L while Mexico provisions stay elevated — that combination would compress ROE toward 22% and re-rate the multiple toward Itaú-anchored 2.1–3.2x book. The verdict would shift to clean Lean Long on a Q2 print with provisions at or below $1.4B and NPL stable to declining, and would shift to Avoid on a Q2 print with provisions at or above $1.6B and a forced coverage-ratio uplift above 15.4%. The decisive variable is one earnings print roughly three months out, so the institutional posture is patient ownership at most, not chasing strength.

What, If Anything, Protects Nu Holdings

1. Moat in One Page

Conclusion: narrow moat. Nu has a measurable and company-specific economic advantage, but it is concentrated in one geography (Brazil), one cost line (unit operating cost), and one cohort behavior (rising ARPAC on a flat cost-to-serve). It does not extend cleanly to Mexico, where Mercado Pago is the only peer with comparable returns, and it has not been tested through a deep Brazilian consumer-credit downturn at current scale. A moat exists; the case for wide moat is not proven.

The 2-3 strongest pieces of evidence: (1) ~14,000 customers per employee versus ~1,200 at Brazilian incumbents, driving a ~$0.80/month cost-to-serve that no listed peer matches; (2) a 30.3% FY2025 ROE earned alongside 27% revenue growth — a combination no other listed LatAm bank achieves, with Inter and PagBank stuck at 14% ROE despite being digital-native; (3) primary-banking-relationship penetration of roughly 58–65% of 1-year-tenured customers, anchoring a deposit franchise funded at 81% of CDI. The 1-2 biggest weaknesses: Inter funds itself at 65.3% of CDI — 160 bps cheaper than Nu — proving the deposit-cost advantage is a function of scale and brand, not digital-bank physics; and the cost-of-risk line at ~19% of average loans has never been tested through a real Brazilian recession at the current $27.7B loan book.

Moat Rating: Narrow moat · Weakest Link: Mexico replication risk vs Mercado Pago

2. Sources of Advantage

Five candidate moat sources are testable against the upstream evidence. Two pass with high confidence, two pass with qualifications, one fails. The point of this table is to be specific about how each one is supposed to work and what would falsify it.

3. Evidence the Moat Works

Six observable data points, each linked to a moat claim and to a way it could be wrong. The point is to test the moat, not to celebrate it.

The ARPAC chart is the single cleanest visualization of the moat working. Monthly revenue per active customer has roughly tripled since 2019 on a cost-to-serve that has stayed essentially flat at $0.80. The $27/month mature-cohort number is what every other cohort could converge to as it ages — that gap is the moat's compounding mechanism, not a forecast.

The peer ROE chart tells the moat story most precisely: Nu earns a 30% ROE despite a worse cost of funding than Inter — meaning the moat is not cost-of-funding per se, it is scale and cross-sell stacked on top of a still-good (not best-in-class) cost-of-funding base. If Nu had Inter's funding cost, ROE would be materially higher; if Inter had Nu's scale, Inter's ROE would presumably also be much higher. The variable that separates them is scale + cohort maturity, which is exactly the moat.

4. Where the Moat Is Weak or Unproven

The honest skeptical reading of the moat case rests on five concerns, each of which would compress earnings power or refute one of the source claims.

5. Moat vs. Competitors

The Competition tab established the peer set. The point of this table is to translate that into moat-relative strength: where each competitor is stronger or weaker than Nu as a moat — not just on financials.

The moat-relative read of the peer bubble: Nu and MELI are alone in the upper-right quadrant where the market prices a genuine moat. Inter has one moat ingredient (funding) and is being priced at 1.4x book — proof that any single ingredient is not enough. ITUB has the legacy moat (corporate + affluent + insurance) and trades at 2.2x book — the market sees those as durable but not growing. Nu is the only listed LatAm bank for which a 5x book multiple makes sense, but the comparison that should keep an analyst honest is to MELI (15x book / 36% ROE) — that is what a wide moat would look like, and Nu is not currently priced there.

6. Durability Under Stress

A moat must survive things it has not yet faced. Six stress cases, with the realistic Nu response and the signal an analyst would watch.

The heatmap is a reading aid, not a model output: numbers are author judgment, anchored on upstream evidence. The two stresses where moat resilience scores lowest are the Mexico/Mercado Pago contest and a deep Brazilian unsecured-credit downturn — both because Nu has not been tested through them. The two where resilience scores highest are incumbents-catch-up (slow erosion, decade-scale) and a single regulatory cap (Nu has navigated the 2023 cap and is reweighting toward installment + secured).

7. Where Nu Holdings Fits

The moat is not uniform across Nu's portfolio. Three observations matter:

1. Brazil consumer-credit and deposits — moat is strongest. This is the franchise that drives 85-90% of revenue and the great majority of profit. Customers-per-FTE, primary-banking penetration, brand favorability, deposit cost relative to CDI, and BCB-tracked share gains are all evidence of a real, measurable advantage here specifically. Within Brazil, the unsecured credit-card book (24.3% share of purchase volume) is where the moat shows up most clearly because that product mix capitalizes the lowest cost-to-serve and the highest revolving-yield combination.

2. Mexico — moat is partial and contested. Nu is the #3 credit-card issuer (6.6M cards, ~17-18% of total Mexican credit-card market as of June 2025) but Mercado Pago is the #1 fintech by MAU and the only competitor that out-grows and out-monetizes Nu in this market. The moat ingredients that travel best to Mexico are brand and unit cost; the ones that travel weakly are the BCB-driven regulatory tailwind (Mexican Sofipo regime is different) and the data-underwriting flywheel (Mexican credit history is shorter and Mercado Pago's e-commerce TPV gives it more underwriting signal in some segments).

3. SME, US, affluent BR — no moat yet. SME (Working Capital launched 2024, Charging Assistant 2025) is early and competes against PAGS and Stone which have multi-year head starts on merchant relationships. US (OCC conditional Jan 2026) is pre-revenue and Nu will be a sub-scale challenger in a market dominated by Chime, SoFi, and the largest US banks. Affluent Brazil banking is Itaú Uniclass and Bradesco Prime territory and Nu's Ultravioleta tier is small. These are call-options on the future moat, not part of the current moat.

8. What to Watch

Eight observable signals will tell you whether the moat is widening, holding, or narrowing — in roughly the order they tend to move.

The first moat signal to watch is Nu's cost of funding as a percentage of CDI. It moves before earnings, before NPL, before customer count; it sits at the structural intersection of brand, primary banking, and deposit franchise; and Inter's 65.3% disclosure proves the ceiling is set by competition, not by digital-bank physics. A reading that stays in the 75-85% band keeps the moat intact. A reading that breaks 88% in a quarter where Selic is flat is the metric to act on — that is the deposit franchise weakening, and the deposit franchise is the spine of every other moat ingredient.

Financial Shenanigans — Nu Holdings Ltd. (NU)

Nu Holdings clears the high-conviction forensic tests but does not clear the high-conviction governance tests. There is no restatement, no admitted control weakness, no auditor resignation, no live regulatory enforcement, and earnings convert to cash at $1.22 of CFO per $1 of net income — a clean bill from the income-versus-cash test. The risk profile sits in three softer places: a founder-CEO who controls 74% of the voting power under the NYSE "controlled company" exemption, three parallel reporting frameworks (IFRS, FX Neutral, Adjusted Net Income, and now a new Managerial P&L that carries only KPMG "limited assurance"), and a 77% one-year jump in the loan book against an expected-credit-loss coverage ratio that management chose to hold flat at 15.4%. The single data point that would most change the grade is the 90+ delinquency reading and coverage ratio in the next two quarters: a coverage uplift would confirm provisioning discipline, a coverage drop would convert a yellow flag to red.

1. The Forensic Verdict

The forensic verdict is Watch (30/100). Earnings quality and cash conversion are strong, balance-sheet "soft asset" intensity is low for a bank scaling at 50%+, and there is no external evidence of accounting misconduct, restatement, or auditor concern. What keeps the score above the "Clean" band is a layered metric framework introduced at the same time growth started normalizing, plus founder control that removes the usual independent check on aggressive reporting.

Scorecard — 13 shenanigan categories

2. Breeding Ground

The structural conditions are mixed. The audit committee chair is a former PwC Brazil audit partner with a Brazilian CPA, the audit committee is fully independent, and the FY2025 20-F explicitly states "we have not been required to prepare an accounting restatement at any time." But the company is incorporated in the Cayman Islands, operates almost entirely in Brazil, files as a foreign private issuer (no Form 4 insider trading visibility), uses the NYSE "controlled company" exemption, and concentrates voting power and director-nomination rights in the founder-CEO.

The breeding ground is dampening, not amplifying. The architecture is what you would expect of a founder-controlled LatAm fintech that took the cleanest possible path through US capital markets: Cayman holding company, IFRS reporter, KPMG auditor, audit-committee chair with deep audit credentials, no restatement history, no compensation metric obviously tied to a manipulable non-IFRS number. The single concentrated-control risk is real but is a governance choice the market is already pricing — it is not, on its own, evidence that the numbers are stretched.

3. Earnings Quality

Earnings quality looks strong but has two pressure points: heavy capitalization of intangible assets and a balance-sheet expected-credit-loss coverage ratio that is unchanged through aggressive loan growth.

Income statement is matched by the balance sheet

Revenue grew 37% in FY2025 against a 50% jump in total assets, a 49% increase in credit card receivables, and a 77% jump in loans to customers. The receivables-to-revenue relationship is not a "DSO blowing out" story — both sides moved in step, and the gross loan portfolio remains short-duration (credit cards plus consumer loans). The forensic ratio that matters more for a bank is ECL coverage, which management held at exactly 15.4% in FY2024 and FY2025.

Reserve coverage held flat through a 56% portfolio jump

ECL coverage stable at 15.4% is the single most important forensic statement in this filing. Management is explicit that the stability "reflects that the increase in the ECL was primarily driven by the growth of our total portfolio and remained aligned with its underlying credit quality." The 90+ NPL ratio in Brazil disclosed at 6.6% is consistent with that. But under IFRS 9, a stable coverage ratio while the loan book grows 77% is only defensible if the new vintages carry the same expected-loss profile as the seasoned book. If the FY2026 90+ NPL ticks above 7% without a coverage uplift, the ratio will look mechanical, not analytical.

Capitalization of soft costs is rising

Intangibles jumped 73% YoY to $602M and "Other assets" 112% YoY to $1.40B. The MD&A explains the latter includes deferred card-issuance expenses amortized over time. Capex per cash-flow statement reads $7M (property, plant & equipment only); capex per MD&A reads $340.8M because it includes intangible-asset development. The line item disclosure is honest, but a reader who anchors on the cash-flow statement understates investment intensity by 47x. Combined goodwill plus intangibles is still only 1.4% of total assets, well below software peers, but the gap between depreciation/amortization ($98M FY25) and full capex ($341M) means there is a $243M annual book accumulation flowing through the balance sheet rather than the P&L.

Non-operating "Other income" is small and stable

"Other income" is small (3.4% of pretax FY25) and is disclosed as FX gains on foreign transactions and monetary adjustments on recoverable taxes — these are mechanical, not one-time gain manufacturing. Clean.

4. Cash Flow Quality

Cash flow quality is the single strongest part of the forensic picture. Earnings convert to cash, but the conversion mechanism deserves to be named: this is a bank, not a SaaS company, and CFO strength is the natural by-product of deposit funding.

CFO consistently exceeds net income

Three years in a row CFO and FCF have landed within $30M of each other at roughly 1.22x net income. The non-cash adds bridging the two are dominated by ECL provisions ($4.7B added back in FY25 reconciliation), interest accruals, and deferred tax movements — exactly what you would expect from a fast-growing IFRS bank.

Accrual ratio is clean

The accrual ratio has been negative — CFO exceeds NI — for four consecutive years. The FY2021 spike is the pre-IPO ramp where loan-book growth ($-$2.9B of operating cash use) outran net income; that pattern reversed once deposits matched receivables.

Source of CFO: deposit growth, not working-capital tricks

Deposits grew $13.0B in FY2025 — that is the dominant inflow inside the $3.5B CFO line. Per IFRS, deposit growth is correctly classified as an operating cash inflow for a bank. There is no securitization, no recourse-factoring program, no supplier finance, and no "off-balance-sheet" receivable sale disclosed. The CFO line is real, but it is structurally dependent on deposit growth keeping pace with loan growth. Borrowings and financing also tripled to $4.4B, suggesting management is starting to lean on wholesale funding to top up retail deposits.

Capex picture depends on which line you read

The cash-flow statement reports capex as $7M FY25 because it tracks only property, plant & equipment additions. The MD&A reports total capex (including capitalized intangibles) of $340.8M — a 47x gap with what most data feeds will surface. Depreciation/amortization is $98M, so the full-capex / D&A ratio is roughly 3.5x. This is a disclosed convention, not deception, but it shows up in the data files in a way that materially understates investment intensity. Any forensic check that uses "capex per cash-flow statement" as a productivity proxy on this name will be wrong by an order of magnitude.

5. Metric Hygiene

This is where the forensic picture earns its yellow flags. Nu reports four distinct earnings frameworks side by side: GAAP/IFRS, FX Neutral, Adjusted Net Income, and now the new Managerial P&L. Each is internally consistent. The risk is that the headline number an investor anchors on changes from quarter to quarter, and that the framework drift is doing the work of accounting choices that would otherwise be debated more openly.

Non-IFRS gap is closing — that helps

The Adjusted-to-GAAP gap has compressed from 16.1% in FY2023 to 7.1% in FY2025 as the SBC add-back has stayed flat in dollars while GAAP earnings tripled. This is a positive forensic signal — management's preferred number is converging on the audited number, not diverging.

The Managerial P&L is the new red flag

The Q1 FY2026 6-K is the first reporting period in which the new "Managerial P&L" framework is used as the primary disclosure. The release states KPMG performed a limited assurance engagement on the compilation process — meaning KPMG concluded "nothing came to its attention" suggesting impropriety, which is a deliberately lower bar than the reasonable assurance of a full audit opinion. Limited assurance is a recognized professional standard, not a flaw, but it is the right framework to use to bridge audited IFRS results into a presentational re-cut — not the right framework to use as the headline number an investor sees first.

6. What to Underwrite Next

This work belongs in the diligence-overlay bucket, not the thesis-breaker bucket. The earnings, cash flow, and balance sheet pass the high-conviction tests. The governance, presentation, and reserve-policy items are real but priced. A position-size haircut of 5-10% versus what the income-statement quality alone would suggest is reasonable; a valuation haircut is not.

Five specific items to monitor

Practical use

Treat this as a position-sizing limiter, not a thesis breaker. The accounting plumbing is sound, earnings convert to cash, and there is no external evidence of misconduct, restatement, or auditor concern. The risk is structural: a founder who controls the vote, a new presentational framework that carries only limited assurance, and a stable headline ECL coverage ratio that the next two quarters of credit data will validate or invalidate. Underwrite with a 5-10% sizing discount versus what the earnings quality alone would dictate, monitor the Q2 and Q3 FY2026 90+ NPL print and the IFRS-vs-Managerial-P&L bridge, and revisit the grade when the Managerial P&L has either three quarters of consistent definitions or a definition change that invites questions.

The People Running Nu Holdings

Governance grade: B. Founder-controlled, deeply aligned, and stacked with top-tier independent directors — but the dual-class structure gives David Vélez 74% of the vote on far less than 20% of the economics, and the controlled-company exemption blunts the protections minority holders would otherwise rely on. Trust is high on capability and alignment; structural power asymmetry is the offset.

Governance Grade: B

The People Running This Company

Nine people decide what Nu does. Two of them — David Vélez and Cristina Junqueira — founded the company together in 2013, still run it, and between them control about 84% of the votes. Everyone else is a hired executive or an independent director.

The team passes the threshold question — does it look like a real bank? Yes. Lago (CFO) and Fragelli (CRO) bring orthodox global banking pedigrees (Credit Suisse, HSBC). The hires of Roberto Campos Neto (former Brazilian central bank president, architect of Pix) and Eric Young (Google/Snap/Amazon engineering) in 2025 directly address Nu's two scaling risks: regulatory navigation as it expands across LatAm and pursues a US charter, and platform engineering as it crosses 130M customers. Campos Neto's hire is the most consequential governance event of the past year — it is rare for a sitting G20 central banker to join a private bank within a year of leaving office, and it materially strengthens Nu's standing with regulators in Mexico, Colombia, and the US.

The Lahrech departure (May 2025) and Vélez resuming direct command is the one watch-item. Founders who reclaim operational control after a President exits sometimes signal either decisiveness or stretched bandwidth — Nu's Q1-Q4'25 execution suggests the former, but the hiring cadence (CTO, CDO, CPO, Vice-Chairman, plus a new CPO in May 2026) shows succession is being built around Vélez rather than under him.

What They Get Paid

Nu discloses aggregate, not individual, compensation — it is a Cayman-domiciled foreign private issuer and is not required to. The headline number is $91.3M in aggregate FY25 compensation for the board and key management, of which the overwhelming majority is share-based.

The 84% equity weighting is unusually high for a regulated bank and is driven by the Contingent Share Award (CSA) granted to Vélez in 2021: a one-time long-term grant worth ~1% of Nu's share capital, but only earned if the share price hits specific thresholds (with the first threshold around $18.69 — roughly 3x the IPO price). Vélez has publicly committed via the Giving Pledge to donate the entirety of any CSA proceeds to a family philanthropy focused on LatAm youth — so the program functions more as a control-and-retention device than personal wealth creation.

Is pay sensible? $91.3M in aggregate is roughly 0.6% of FY25 revenue ($15.8B) and 3.1% of net income ($2.9B). For a $63B-market-cap bank with 33% ROE and 10,000 employees, this is in line with — and arguably below — what a large US or European bank pays its top tier. The Compensation Committee is chaired by Jacqueline Reses (ex-Square Financial Services, CEO of Lead Bank) and includes Doug Leone (Sequoia) and Luis Moreno (former IDB president) — three people who have seen elite executive comp at scale and are not pushovers. The clawback policy was adopted in October 2023 and is current with SEC rules. What you do not get with a Cayman FPI: individual NEO disclosure, a true Pay-vs-Performance table, or a binding say-on-pay vote. That is a structural protection minority shareholders simply do not have.

Are They Aligned?

This is the strongest section of the file. By every meaningful test of alignment — founder economic stake, co-founder stake, equity-weighted comp, restraint on dilution, absence of related-party self-dealing — Nu is at the top of its peer group. The asymmetry is in control, not in alignment.

Vélez's Class B shares carry 20 votes each versus 1 for Class A. He owns about 18.6% of the company economically but controls 74.4% of the votes. That gap — 56 percentage points — is the single biggest governance risk in the name. It means:

• Strategic decisions, board composition, and amendments to the M&A all run through Vélez alone.
• Nu invokes the NYSE controlled-company exemption, electing not to be required to have a majority-independent board or a fully independent comp committee (Nu chooses to anyway, but it could stop).
• The Shareholder's Agreement gives Vélez personal consent rights over major actions (M&A, dividends, share issuances over 20% of equity, related-party transactions, etc.) so long as he holds 10%+ of votes. He is not just the largest holder — he is a one-man second board.

Insider activity. Insider selling is minimal and routine. The one discretionary sale in the past year was Cristina Junqueira selling 300,000 shares at $14.81 on March 23, 2026 — a $4.4M transaction that reduced her direct holdings by 11.48% but is trivial against her ~$1.7B total stake (she still controls ~131M shares through trusts and LLCs). The CRO had a tax-withholding disposition in April 2026 — these are mandatory, not discretionary. There has been no insider open-market buying disclosed, but founder-controlled companies rarely show it.

Dilution. Nu's equity plans authorize up to 5% of fully diluted shares per year. At ~4.86B shares outstanding, that is a notional ceiling of ~243M shares — meaningful. Actual outstanding RSUs + options sit at 74.9M (21.8M options + 53.1M RSUs), about 1.5% of shares. So the live overhang is small; the cap is high. SBC ran roughly $400M in FY24-25, real but absorbed by 33% ROE.

Related-party transactions. The 20-F discloses only ordinary credit cards and personal loans to executives and directors, on the same terms offered to other customers, and a de minimis "Others" line of $1,795 in FY24. No promoter loans, no acquisitions from affiliates, no real-estate side deals. This is exceptionally clean for a controlled company.

Capital allocation. Nu does not pay a dividend and has done no buybacks — every dollar of $2.9B in FY25 net income is being reinvested in customer growth, credit book expansion, and the US launch. Given 33% ROE, this is the right call. A buyback or special dividend would actually be a yellow flag at this stage.

Skin-in-the-game score: 9/10. Founder and co-founder hold ~$13B in stock between them, executive comp is overwhelmingly equity, the CEO's largest grant only vests on share-price thresholds, and pledged philanthropy on those proceeds removes the personal-enrichment angle. One point off for the dual-class structure, which converts excellent alignment into excellent control — minority holders ride with him, but cannot challenge him.

Board Quality

Seven of nine directors are formally independent. More importantly, the substantive quality of the board is unusually high for an emerging-markets bank. Each independent director brings either operating experience at a globally-scaled platform, deep banking/audit expertise, or regulatory standing.

Audit & Risk Committee — chaired by Rogério Calderón Peres, a former PwC Brazil audit partner and CFO at Bunge Brasil, Unibanco/Itaú Unibanco, and HSBC Brasil. He is one of the strongest bank-audit-chairs in LatAm capital markets. Joined by Anita Sands (ServiceNow board) and Thuan Pham (former Uber/Coupang CTO) — a balanced mix of financial-reporting and tech-risk expertise. This is a real committee, not a rubber stamp.

Compensation & People Committee — chaired by Jacqueline Reses, ex-Square Financial Services and currently CEO of Lead Bank. Includes Doug Leone (Sequoia Global MP 2012-22), Luis Moreno (15-year IDB president), and David Marcus (ex-PayPal President, ex-Meta payments, now Lightspark CEO). On paper this committee could push back on any pay or equity proposal — whether it actually does is unverifiable given the lack of individual disclosure.

Where the board is thin: real cybersecurity-specialist depth (Sands is the closest); a director with hands-on consumer-credit underwriting/loss-given-default experience separate from management; and an independent voice from Mexico or Colombia (the two growth markets). Marcus's crypto/Bitcoin/stablecoin focus at Lightspark is a watch-item if Nu accelerates NuCrypto.

The structural caveat. Substantive independence ≠ structural independence. Vélez retains the right under the Shareholder's Agreement to nominate up to 5 directors (a majority) for as long as he owns 40%+ of votes. He currently nominates fewer than that. But if a real disagreement arose between the independent directors and Vélez over a strategic decision, Vélez wins — he could replace them. The current board is excellent because Vélez chooses excellent people. There is no governance mechanism preventing him from choosing differently.

The Verdict

Overall Governance Grade: B

Letter grade: B.

The bet on Nu is, in governance terms, a bet on David Vélez's judgment and stewardship. The infrastructure around him — board, audit chair, CRO, CFO, the Campos Neto hire — is among the best available, and his own economic interests are vastly more aligned with the public float than the typical founder-CEO. The structural risk is that none of the protections are binding against him. For a fundamentally sound, profitable, founder-driven bank growing at 40%+, that is an acceptable trade — but it is a trade, not a free option.

How the Story Changed

Nu Holdings has been telling essentially the same headline story since David Vélez co-founded the company in 2013 and rang the IPO bell in December 2021 — disrupt Latin American banking, compound a low-cost deposit franchise, monetize the same customer over time. Across the most recent seven filings, the language shifted considerably: "scale efficiently with discipline" (mid-2025) gave way to "AI-first" and "rebuilding banking around AI" (late-2025/2026), Mexico moved from launch-mode to break-even, and a brand-new US national bank charter appeared in the narrative for the first time. Bad news has been rare and pre-empted (PIX Financing slowdown, Nucoins repositioning, Q1'26 ECL spike were each labelled "expected seasonality" or "prudence"), and credibility has improved — but the deck is now stretched across three geographies, two consumer segments, an AI rebuild, and a US optionality bet, all at once.

1. The Narrative Arc

The arc has three distinct chapters. 2013–2020 was a single-product disruptor scaling on word-of-mouth. 2021–2024 was the public-company "platform" phase — broadening the product mix from credit cards into deposits, payments, lending, investments, insurance. 2025 onward is the AI-and-geographic-optionality phase: foundation models in production, a US charter, a strategic investment in Tyme (African digital bank), and Mexico finally turning the corner. The same CEO has narrated all three chapters, which is unusual for a company that has gone from ~3M customers to ~135M in twelve years.

2. What Management Emphasized — and Then Stopped Emphasizing

Three patterns jump out of the heatmap:

• Three themes have only intensified — Mexico expansion, secured lending, ARPAC monetization. These are the load-bearing pillars of the current bull case.
• Two themes appeared from nowhere — "AI-first" (Q3 2025) and the US bank charter (Q4 2025 onward). Both are now top-of-page in management's framing, but they were not mentioned at all twelve months earlier. The AI-first rhetoric quieted slightly in Q4 2025 (the year-end report) then re-emphasized in Q1 2026 ("not adding AI to banking, we are rebuilding banking around AI").
• Two themes were quietly dropped — PIX Financing expansion (after Q3 2024's "intentionally slowed pace of eligibility expansion") and the Nucoins loyalty program (which absorbed a $40M one-off marketing expense in September 2024 and was then never trumpeted again). Neither was framed as a failure; both simply stopped being mentioned.

3. Risk Evolution

Three risks have risen in the FY2025 risk-factor disclosure:

• AI risk is newly material. Last year's 20-F barely mentioned generative AI; the FY2025 filing now lists "use and provision of solutions powered by artificial intelligence could lead to operational or reputational damage, competitive harm, legal and regulatory risk and additional costs" inside the top-tier risk summary. That's a direct concession that the "AI-first" pitch from the call carries real downside.
• Brazil political risk has been promoted ahead of the October 2026 presidential elections. The FY2025 filing explicitly flags election uncertainty as something that "may further increase volatility in the market price of securities."
• International expansion now carries the US too. The "international expansion may not be successful" risk used to mean Mexico and Colombia; in the FY2025 filing, it now covers the OCC-approved US bank too.

Notably, two risks that should be discussed more visibly remain at constant intensity: Vélez's 74.4% voting power (which structurally limits any activist or board response if execution slips), and the IT-systems-at-scale risk (now arguably understated given the AI rebuild's dependency on third-party foundation models and cloud capacity).

4. How They Handled Bad News

Nu has not had a crisis in the period covered — no scandal, no SEC enforcement, no founder resignation, no asset write-down. Instead, the test of management's communication style comes from three soft setbacks:

The pattern is consistent: pre-emptively frame the bad news as discipline, then never re-litigate it. This works when the underlying business is compounding — and it has — but it leaves no audit trail for investors who want to know whether the original concern was right. PIX Financing's "more data to ensure credit models remain resilient" caveat was never explicitly answered; Nucoins was never tested against a revenue counterfactual.

"In Q1'26, we delivered our first quarter of IFRS profitability ahead of internal plan, ARPAC has nearly doubled, even as we have onboarded millions of newer, less mature customers." — David Vélez, Q1 2026

This is the template: lead with the milestone, follow with the volume metric that supports it, and never directly engage the EPS-miss-versus-consensus framing that the sell-side actually moved on. It is effective communication. It is also non-falsifiable in the short term.

5. Guidance Track Record

Nu rarely issues hard numerical guidance; the "promises" worth scoring are the directional commitments that the market priced and capital allocators acted on.

Credibility score: 8 / 10. Of eleven valuation-relevant commitments, seven have been delivered or are tracking ahead of plan. Two — PIX Financing re-acceleration and the Nucoins loyalty payoff — were quietly dropped rather than failed-and-acknowledged. Two more (AI-first and US optionality) are aspirational and unscoreable today. The half-point markdowns are for the style of how soft setbacks were communicated (pre-empted, then never re-litigated), the broadening of new commitments without first closing out old ones, and the structural fact that Vélez owns 74.4% of the voting power, which means external accountability for any future drift will be limited.

6. What the Story Is Now

The current story is the cleanest version Nu has ever told: the largest private bank in Brazil by customer count is now a profitable, AI-rebuilding, three-country platform with a US optionality bet attached. The core compounding engine — customer acquisition, deposit franchise, ARPAC monetization, falling cost-to-serve — has done what management said it would do across every public quarter, though the marquee external long anchor changed in 2025 (Berkshire fully exited Q1 2025, disclosed via 13F/A Aug 14, 2025).

What has been de-risked since 2021: Brazil scale (largest private bank), GAAP profitability (achieved 2023, $2.9B in 2025), Mexico monetization (break-even Q1 2026), secured-lending diversification (33% of portfolio now non-credit-card), and the efficiency-ratio thesis (~27% headed below 20%). What remains stretched: the AI-first claim is real enough to be in production but cannot yet be cross-checked against margin expansion, and US-charter optionality is being scored as "bounded downside" by a team that has never run a US national bank. The Tyme strategic-investment call-options that appeared in the Q1 2026 financials are not yet a story management is publicly building on — worth tracking.

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.

Web Research

The web tells a tighter story than the filings alone. Nu Holdings just printed record Q1 2026 revenue ($5.32B) but missed EPS ($0.18 vs. $0.20) on a 33% sequential jump in credit-loss allowance to $1.79B — and the stock fell roughly 7% on May 15, 2026. Layered on top: a sitting CEO who sold $300M+ of stock in August 2025, the unexpected May 2025 departure of the President/COO running the credit book, a co-founder relocating to the U.S. to launch a de novo national bank, and a former Brazilian Central Bank president parachuted onto the board.

1. The Bottom Line from the Web

The single most important web finding: Nu's reported growth narrative is intact, but the cost — credit risk and management instability — is rising faster than the filings will tell you on first read. Provisions jumped 33% QoQ in Q1 2026, the President/COO who built the credit book left abruptly in May 2025, CEO David Vélez monetized 33M shares in August 2025, and Cristina Junqueira (co-founder, now running U.S. expansion) sold $4.4M of stock in March 2026. Analyst consensus remains a Buy ($15.81 average target, 16 analysts) — but Zacks downgraded to Hold on March 24, 2026 and the stock is down ~21% over the past 90 days to ~$12, near the 52-week low of $11.71.

2. What Matters Most

Finding 1 — Q1 2026 EPS miss masked by record revenue, driven by credit provisions

The 90+ day NPL ticked down 10 bps to 6.5% (below the Q3 2024 peak of 7.0%), so the deterioration is leading-indicator only — but the size of the provision jump unnerved investors fixated on the EPS line.

Finding 2 — President/COO Youssef Lahrech departed in May 2025; CEO absorbed the role

This is the highest-leverage governance-vs.-results pairing on the page: the executive responsible for credit-model design left just before credit losses spiked, and the CEO — not a new credit specialist — now owns the function.

Finding 3 — CEO sold 33 million shares in August 2025

Finding 4 — Nubank, N.A. — conditional U.S. national bank charter

Finding 5 — Campos Neto appointment raises governance/independence questions

Finding 6 — Co-founder Junqueira sold ~$4.4M; insider holdings now ~1.20%

On March 23, 2026, Cristina Junqueira (US CEO & Chief Growth Officer; co-founder) sold 300,000 shares at an average of $14.81 — $4.44M total — reducing her position by ~11.5%. Per InsiderScreener, four insider transactions in the prior 90 days were all sales totaling $8.89M; aggregate insider ownership now stands at ~1.20% (excluding Vélez's controlled Rua California stake). Source: insiderscreener.com, marketbeat.com.

Finding 7 — Berkshire Hathaway trimmed NU 19% in Q3 2024

Per Motley Fool (Jan 11, 2025), Berkshire Hathaway sold 19% of its Nu position in Q3 2024. Nu still represents only ~0.3% of Berkshire's equity portfolio, so the trim is small in absolute terms, but a Buffett trim of a high-multiple Brazilian bank during a stronger BRL period drew attention. Source: fool.com.

Finding 8 — TikTok's Brazil fintech move is a credible competitive overhang

Finding 9 — Tyme Group $150M Series D — quiet but strategic

Nu led a $150M investment in Tyme Group (digital bank in South Africa and the Philippines) in December 2024. The deal signals optionality on emerging-markets digital banking outside LatAm but came under investor scrutiny given Tyme's profile and Nu's already-stretched US/Mexico expansion. Source: gurufocus.com.

Finding 10 — Velez's 2021 CSA termination (historical context)

In November 2022, Vélez voluntarily terminated the 2021 Contingent Share Award — saving Nu an estimated $356M over seven years. A pro-shareholder gesture that should be remembered alongside the August 2025 $500M secondary sale: both data points belong in any "founder alignment" assessment. Source: international.nubank.com.br.

3. Recent News Timeline

4. What the Specialists Asked

5. Governance and People Signals

Board / Management changes 2024–2026 (key entries):

Pattern read: Five material people events in 12 months — heavy churn at the top during a period when credit metrics are softening. Vélez is now personally responsible for the credit book (Lahrech gap), the US expansion (Junqueira on the ground), and overall strategy. Concentration of execution risk on one individual is elevated.

6. Industry Context

Mexico: Profit pool $40B+; banking system structurally under-penetrated (less than half of adults hold a formal credit product). Mercado Pago is Nu's closest scaled fintech competitor in both Brazil and Mexico. Nu Mexico now #1 issuer of new credit cards.

United States: Nu received OCC conditional approval (Jan 29, 2026) for Nubank, N.A. — a de novo national bank in McLean, VA, capitalized within 12 months and operational within 18 months. Reportedly targeting the Hispanic population (~20% of US residents) and four geographic hubs (Miami, SF Bay Area, Northern Virginia, North Carolina Research Triangle). Stiff competition expected; market is dominated by incumbents and well-capitalized fintechs (SoFi, etc.).

Competitive overhangs of note: (i) TikTok's April 2026 push into Brazil fintech; (ii) Itaú's mobile-first "Itaú 1" initiative; (iii) Mercado Pago's continued credit portfolio expansion across LatAm; (iv) C6 Bank's JPMorgan-backed funding base; (v) Inter's deposit growth.

Where We Disagree With the Market

Consensus has decided the May 14 reset already priced the Q1 credit-cost crack, and is now positioned for a "buy the dip" recovery into the Q2 print: 18 of 21 ratings sit at Buy with the median 12-month target clustered between $17.60 and $19.87 against a $12.07 spot. The sharpest disagreement: the market treats Nu's 81%-of-CDI funding cost as a structural moat, but Inter & Co's FY2025 20-F discloses a smaller digital peer funded at 65.3% of CDI — a 160 bps disclosure-level gap that means the deposit-cost edge is narrower than the bull case (and most sell-side models) embeds. Two further gaps follow: consensus reads the 15.4% expected-credit-loss coverage held exactly flat through 56% loan-book growth as analytical reserve discipline, while the forensics evidence reads it as a denominator artifact that has to break in Q2; and consensus models Mexico as a 5-year-lagged Brazil cohort replica, while the competitive evidence shows Mercado Pago — which already earns 36% ROE off captive e-commerce traffic Nu cannot reach — is the load-bearing constraint on that thesis. All three resolve on the August 13–14 Q2 print or the disclosures it forces.

Variant Perception Scorecard

Reading the score. Consensus clarity scores high (78) because there is a tight cluster of ratings — 18 Buys / 2 Holds / 1 Sell with PTs concentrated $17.60–$19.87 — and a clean "buy the dip on credit normalization" narrative across the most recent CICC, UBS, Susquehanna, Morgan Stanley notes. Evidence strength scores 68 because two of the three disagreements rest on disclosure-level peer data (Inter 20-F cost of funding; Mercado Pago Mexico monetization) and one rests on the exact-decimal stability of the 15.4% coverage ratio across two fiscal years — strong but not unfalsifiable. Variant strength sits at 62, not higher, because we are not betting against the franchise: we are pricing risks the consensus appears to underweight, on a name where the bull engineering case (30% ROE, 27% revenue growth, $3.5B FCF) is largely correct.

Consensus Map

The Disagreement Ledger

Disagreement 1 — Deposit-cost moat is narrower than priced. Consensus would say: Nu's 81% of CDI funding is structurally cheaper than the incumbents' 100%+ and supports a multi-year NIM premium; the deposit base ran past $40B because the franchise is sticky. Our evidence says: Inter — smaller, digital, Brazil-only, also Open-Finance-enabled — discloses 65.3% of CDI in the same period; Nu's ratio has crept up from the high-70s as the book scaled; and long-term debt tripled to $4.4B with $2.3B of net wholesale inflow exactly when management is talking about deposit strength. If we are right, the market has to concede that 100-160 bps of the assumed NIM premium is competitive headroom Nu could be giving up rather than a structural floor. The cleanest disconfirming signal: Nu's cost of funding drifting back toward 75-78% of CDI without a corresponding Selic move, with wholesale-borrowing share flat or down.

Disagreement 2 — 15.4% ECL coverage is mechanical, not analytical. Consensus would say: management held coverage flat because the underlying credit profile is unchanged; the Q1 spike was Mexico seasoning + mix and risk-adjusted NIM will rebuild to 10.5% in H2 2026. Our evidence says: a coverage ratio held to one decimal across two fiscal years and 56% portfolio growth is exactly the disclosure shape that precedes a forced uplift when a cycle turns; the Q1 2026 provisions jump (+31% QoQ to $1.72B), the 89-bp jump in 15-90 NPL to 5.0%, and Lahrech's departure as the architect of the credit book nine months earlier are coherent signals of a regime change, not a seasonal print. If we are right, Q2 forces a coverage uplift through the P&L exactly when consensus expects stabilization. The disconfirming signal: Q2 provisions ≤$1.6B with 90+ NPL ≤6.6% and coverage stable at 15.4% without strain.

Disagreement 3 — Mexico is a structurally different game. Consensus would say: Mexico has 15M customers, the #1 issuer of new credit cards, and the cohort ARPAC math will work like Brazil with a 3-5 year lag. Our evidence says: Mercado Pago is already 36% ROE / 39% revenue growth in Mexico from a captive e-commerce flywheel Nu cannot replicate from a banking license; Nu's own Q1 2026 management commentary blamed Mexico loan-book seasoning for the provisioning spike, meaning Mexico is paying for itself with credit losses before it pays for itself with ARPAC; and the Moat tab itself concedes the data-network advantage does not transport cleanly across countries. If we are right, Mexico settles at 18-22% ROE rather than replicating Brazil 30% — the difference between a 5x and 3x book multiple as the customer-base mix shifts away from Brazil. Disconfirming signal: Mexico-specific risk-adj NIM disclosure showing positive contribution and improving credit cost across two consecutive prints, with Mercado Pago Mexico ARPAC not pulling further ahead.

Disagreement 4 — Stable-metric headline mask is a governance discount. Consensus would say: $0.80 cost-to-serve, 15.4% ECL coverage, "above 85" NPS, and a stable Managerial P&L bridge are positive operational signals from a high-quality franchise. Our evidence says: each of these numbers is held to a rounding convention that flatters consistency, the Managerial P&L carries only KPMG limited assurance (a deliberately lower bar than reasonable assurance), the CEO monetized $500M of stock in August 2025 with no public 10b5-1 disclosure surfaced in the research, and the President/COO who built the credit book departed five months later. If we are right, the priced governance discount is 0 and the analytical discount should be 5-10% — small on its own, but it compounds the magnitude of any thesis-breaker above. Disconfirming signal: the Managerial P&L bridge to IFRS staying consistent across Q1-Q3 2026 prints with no definition drift, and the cost-to-serve / ECL coverage lines breaking their rounding bands without P&L surprise.

Evidence That Changes the Odds

How This Gets Resolved

What Would Make Us Wrong

Take the deposit-cost disagreement first, because it carries the most evidence and the most downside if we are wrong. Nu over-funds its loan book by $14B ($41.9B deposits versus $27.7B loans) and runs 1.49x cash-and-securities cover on deposits — both well above incumbent ratios. If the 81% of CDI reflects management choosing to over-fund into a high-Selic environment rather than competitive ceiling pricing, then the 16 percentage-point gap to Inter is not headroom Nu is giving up; it is excess liquidity Nu can roll into higher-yielding assets as Selic falls. If two more quarterly prints show Nu's cost of funding drifting down toward 78% of CDI as Selic eases while wholesale-borrowing share stays flat, the variant view dies on the disclosure tape — and the bull base case of NIM expansion on funding-cost compression becomes the more probable path.

Take the ECL coverage disagreement next. The Bull case for 15.4% being analytical is genuinely defensible: under IFRS 9, expected credit loss is a forward-looking weighted-probability calculation, and management's stated position is that the 56% portfolio growth was funded by new-vintage cohorts with credit quality identical to seasoned cohorts. If 90+ NPL has already peaked at the Q3 2024 7.0% reading (as Q1 2026's 6.5% suggests), then a flat coverage ratio against a falling delinquency rate is not mechanical — it is reserving stability despite improving credit. The cleanest signal that we are wrong: 90+ NPL declining below 6.0% in Q2 with no coverage uplift. That would prove the 15.4% line is an outcome, not a denominator.

The Mexico disagreement is the most contested. The variant view rests on Mercado Pago's reported metrics, but those numbers aggregate seven LatAm geographies and Nu has actually overtaken Mercado Pago as the #1 issuer of new credit cards in Mexico (6.6M holders, June 2025). If Mexico ROE in years 4-5 from launch matches Brazil's year 4-5 — and the underlying credit data so far supports it — Mexico becomes the second engine the bull case requires and the variant compression assumption (18-22% Mexico ROE) is too pessimistic. The cleanest disconfirming signal: standalone Mexico ROE disclosed at 22%+ in any FY2026 quarter.

And finally, on the governance overlay: Vélez's Aug 2025 $500M sale could be a tax-loss-harvesting structure tied to the 2026 Mexican CNBV license-approval timeline, his $356M voluntary 2022 CSA giveback is a counter-signal that few founders have ever matched, and the Managerial P&L bridge to IFRS has been published cleanly in Q1 2026 with no definition drift in the line items the reader cares about. If that bridge stays consistent for three more prints and the cost-of-funding ratio holds, the governance discount we are flagging may have been priced wrong on the negative side.

The first thing to watch is the Q2 2026 print the week of August 13-14, 2026 — specifically whether provisions for credit losses come in at or below $1.6B with 90+ NPL stable to declining and ECL coverage held at 15.4% without strain. That single print resolves more of the variant view than the next nine months of news flow combined.

Portfolio Implementation Verdict

Liquidity is not the binding constraint. NU trades roughly $501M of value per day against a $59B market cap; a five-day fill at 20% of average daily volume absorbs $431M before becoming the market. The tape setup, however, is bearish — price sits 22% below the 200-day moving average, a death cross printed on April 15, 2026 (one month ago), and the stock is parked at the 2nd percentile of its 52-week range with RSI deep in oversold territory.

Price Snapshot

The Critical Chart — Price vs 50/200 SMA

Caption: Price is below the 200-day moving average (currently $15.46) by 22%. The regime since the February 2026 high at $18.83 has flipped from uptrend to confirmed downtrend; the breach of the 200d in March was followed by a failed reclaim in April and a fresh leg lower into May.

Relative Performance

Momentum — RSI and MACD (18 months)

Caption: RSI at 24 is the deepest oversold reading in the chart and historically marks short-covering bounce zones — but the MACD histogram has just rolled back into negative territory after a brief mid-April attempt to base, so any bounce starts from a downward-trending momentum backdrop. Near term: a tactical bounce is plausible from this RSI level, but the structural read (sub-signal MACD widening lower again) argues for waiting on momentum confirmation before adding.

Volume, Volatility, and Sponsorship

The most institutionally-meaningful spike of the past 18 months is February 21, 2025 — 164 million shares (4.1x average) on a -19% day. That is forced selling, not opinion-shifting, and it left a price wound that the recent rally only partially repaired before failing again. The volume profile of the current decline (March–May 2026) shows 50-day average volume rising as price falls — the classic distribution signature.

Caption: Realized vol at 33.6% sits exactly at the long-run 20th-percentile threshold — this is not a panic decline, it is an orderly distribution. That is significant: drawdowns that happen at calm vol typically reflect supply/demand imbalance rather than acute risk-off, and they tend to bottom only when one side capitulates with a vol expansion (something we have not yet seen).

Institutional Liquidity Panel

Note that 20-day ADV ($501M) is meaningfully below 60-day ADV ($722M) — recent participation has cooled despite the price decline, consistent with passive-flow-driven downside rather than fresh long demand.

Fund-capacity table (5-trading-day fill)

At 20% ADV — a realistic ceiling for an active manager not wanting to print on the tape — a 5% portfolio weight is implementable for funds up to roughly $8.6B AUM in five trading days; a 2% weight scales to $21.6B. At a more conservative 10% participation, those numbers halve to $4.3B and $10.8B. Funds materially above $10B AUM that want a meaningful (5%+) NU position will need multi-week scaling.

Liquidation runway

The median 60-day intraday range is 1.28%, well under the 2% impact-cost threshold — execution friction is modest by mega-cap standards, and a fund holding 0.5% of issuer can fully exit in a week at 20% participation. A 1% issuer-level position requires 2 weeks; a 2% position takes a month. Realistic institutional clip: positions up to 0.5% of market cap (roughly $296M) are clean to enter and exit; anything above 1% becomes a multi-week project that risks moving the tape on news.

Bottom line on capacity: at 20% ADV, the largest position that clears in 5 days is 0.5% of market cap ($296M) — at 10% ADV, the same threshold drops to roughly nil. This stock supports any fund at normal sizing; only multi-billion concentrated positions force staged execution.

Technical Scorecard and Stance

Stance: bearish setup on a 3-to-6 month horizon. Five of six dimensions score negative; only volatility is neutral, and that itself is a warning — calm-vol drawdowns rarely terminate without a panic-vol flush. Two watchpoints define the next regime change: reclaim of the 200-day moving average at $15.46 would invalidate the death cross and put the prior $18.83 high back in play, while a break of the 52-week low at $11.90 opens downside toward the 2024 base in the high-$10s. Liquidity is not the constraint — funds can act when the tape gives them an entry; the posture today is watchlist, with a re-entry trigger above $15.46 or a lower entry only after a high-volume capitulation candle (something the current calm-vol tape has not delivered).