Itaú Unibanco Holding S.A. (NYSE: ITUB) — The Best Bank in Brazil, Priced for Brazil to Keep Going Right

Report date: 2026-07-04 Analyst framework: Competitive-advantage / capital-allocation fundamental research (Greenwald Competition Demystified; Marathon Capital Returns) Price reference: ADR $8.12 (NYSE, 2026-07-02) · 52-wk range ~$5.43–$9.26 · ~12% off the 17-Apr-2026 all-time high · ~11.03bn shares · market cap ~US$85bn · controlling equity R$196.1bn (Dec-2025)

The analytical body (§1–§15 below) carries no buy/sell recommendation and no price target. It discusses valuation only as embedded expectations and scenarios. The single, deliberate exception is the Claude’s Take block immediately below, which is explicitly outside that discipline and is the author’s own independent view.

⚡ Claude’s Take

This block is the author’s own subjective opinion and general information — not investment advice. The analytical sections §1–§15 take no position and carry no price target; that discipline is intact everywhere below this box.

Verdict: HOLD — a genuinely elite bank at a full, un-discounted price. Accumulate only on a Brazil-macro selloff; not a short at any price. Conviction: medium. Directional zone (on the ADR, priced off ~$3.77 book/ADR): I would accumulate below ~1.8× book (~$6.8), find it genuinely attractive below ~1.5× book (~$5.7) — where it traded as recently as early 2025 — treat ~2.0–2.3× book (~$7.5–$8.7) as fair value for a durable low-20s%-ROE franchise, and trim/fade above ~2.5× book (~$9.4). At $8.12 (~2.15× book) you are paying full freight: the stock sits at the ~97th percentile of its own ten-year valuation range on P/E, P/B and P/S simultaneously.

Itaú is, on the numbers, the best-run large bank in Latin America and one of the best emerging-market banks anywhere: a ~23–25% recurring ROE, a sector-best ~39% efficiency ratio (~35% in Brazil), pristine asset quality (90-day NPLs at 1.9%, roughly half the system’s), a fortress balance sheet (CET1 12.3%, BIS ~15%), and a capital-return machine that distributed ~72% of 2025 earnings — including a deliberately front-loaded ~R$24bn extraordinary dividend — while still generating capital. The Gordon math is the punchline that keeps this from being a short: feed the bank’s own ~22.5% ROE and ~14.5% cost of equity into justified P/B and you land at ~2.2–2.45× — almost exactly today’s multiple. The richest-ever price is not a bubble on the fundamentals; it is fair value that leaves no margin of safety. The framing is quality-incumbent-at-full-price / a momentum-extended EM-macro proxy — the factor model is blunt that ITUB is a leveraged bet on Brazil itself (Country:Brazil beta ~1.42, dominating every style factor), trading near multi-year highs after a ~4× five-year run, with a fresh negative 3-month momentum turn.

The reason I stop at HOLD, not accumulate, is that everything the current price capitalizes is at a peak: peak-cycle ROE (no Brazilian bank has held mid-20s ROE through a full downturn), a peak own-history multiple, a JCP tax shield that is being legislated away (guided tax rate jumps from ~9% to ~30%+ in 2026), a mass-retail customer base already lost to Nubank on count and engagement, and — the trapdoor every dollar investor must underwrite — the real. In USD the expected return is roughly a ~5% dividend plus ~4–5% USD book growth ≈ ~9–10%, barely clearing a dollar cost of equity, and a BRL slide toward the ~6/USD of 2024 would erase it. You can be exactly right on the bank in São Paulo and dead money in New York. Bull trigger: Selic eases toward ~12% while managerial ROE holds 22%+ and the real stays firm through the October-2026 election — that turns “fully valued” into “cheap for the quality.” Bear trigger: ROE prints below ~20% for two-plus quarters or the real breaks toward ~6/USD — either exposes ~2.1–2.3× book as a cycle-top. Tag: “The best house in a fully-priced neighborhood — and the currency is the trapdoor in the floor.”

📈 Stock Price Action — Five-Year Event Map

Factual price history — not a recommendation and not a price target.

The arc. Over the trailing five years the ITUB ADR has run a near-4× round trip up: from a COVID/rate-hike low of ~$2.24 (Mar-2021) to an all-time high of ~$9.26 (17-Apr-2026), and it now trades ~$8.12 — roughly 12% off that high and near the top of a 52-week $5.43–$9.26 range. The stock spent 2021–2024 range-bound in the low-$3s to low-$5s, then re-rated violently through 2025–early-2026 on a Brazil rate-cut-and-recovery trade — which is exactly why the valuation now sits at its richest-ever percentile.

Cycle narrative. (1) The stock bottomed in early 2021 as COVID risk and the onset of one of the world’s most aggressive tightening cycles drove capital out of Brazil. (2) It clawed back through 2022 on the global commodity/value rotation and shrugged off the narrow October-2022 Lula win, Selic peaking at 13.75%. (3) Second-half 2023 delivered the first real leg up as the central bank pivoted to cuts and Congress passed a fiscal framework. (4) 2024 was the down year — the real crashed toward ~6.2/USD on fiscal-credibility worries and Selic re-hiking, dragging the dollar ADR down even as BRL earnings held. (5) 2025 was the explosive recovery: the real rallied, Selic topped at 15% with easing on the horizon, earnings set records at a ~23% ROE, and EM flows returned. (6) That momentum carried the ADR to its all-time high by April 2026 as the easing cycle actually began. (7) The most recent ~12% pullback is profit-taking off that high, layered onto management’s own Q1-2026 warning that macro conditions (geopolitics, oil, agribusiness, weaker capital-markets fees) had worsened into an election year. The price move is Fact; the attributed cause is Interpretation.

1. Executive Summary

Itaú Unibanco Holding S.A. (NYSE: ITUB; B3: ITUB4) is the largest bank in Latin America and the largest private-sector bank in Brazil — a universal financial conglomerate formed by the 2008 merger of Banco Itaú and Unibanco, controlled for generations by the Moreira Salles and Egydio de Souza Aranha/Villela families through the IUPAR holding vehicle. It runs a R$1.49 trillion credit book across retail, wholesale (Itaú BBA), and market/corporation segments, overlaid with a deep, capital-light fee-and-insurance franchise (asset management ~R$1.24tn AUM, insurance, cards, acquiring via Rede, a top-tier investment bank), and a Latin-American footprint spanning Chile, Colombia, Uruguay and the Andean region.

The financial record is genuinely elite. FY2025 recurring managerial net income was R$46.8bn (+13.1%) at a 23.4% consolidated recurring ROE (24.6% in Brazil; 24.8% in Q1-2026) — roughly double a typical developed-market bank — earned on a fortress balance sheet (CET1 12.3%, Tier 1 13.8%, BIS ~15%) with the sector’s cleanest asset quality (90-day NPLs 1.9%, ~half the system average) and a best-in-class ~39% efficiency ratio (~35% in Brazil). Earnings are growing roughly twice as fast as the balance sheet — the signature of a mature bank harvesting spread, mix and cost discipline rather than chasing volume. Capital allocation is the standout: management reinvests only mid-single-digit RWA growth that clears its ~23% ROE hurdle, refuses transformative M&A (and actively exited its XP Inc. stake), and returns the surplus — ~72% of 2025 earnings, including a front-loaded ~R$24bn extraordinary distribution — under a newly formalized 12%-CET1-floor remuneration framework with buyback-and-cancellation authority.

Three things temper the quality. First, the moat is narrowing at the mass-retail edge: Nubank overtook Itaú on customer count (112.0m vs 100.3m) and runs nearly double its monthly transaction volume at a structurally lower cost-to-serve; PIX and Open Finance have commoditized payments/float and demolished switching costs. Itaú’s answer — defend the affluent, SME, wholesale and fee pools where its advantages remain decisive — is rational and visible in the segment data (individuals +3.9% vs SME/LatAm high-single/double digits), but it is a defense of a shrinking-share, higher-quality core, not an impregnable position. Second, the earnings are peak-flattered: a ~15% Selic widened liability spreads, and the JCP tax shield held the effective tax rate near ~9% — a shield now being legislated away (2026 guided tax rate 29.5–32.5%). Third, the stock is a Brazil-macro proxy (Country:Brazil factor beta ~1.42): its dollar return is dominated by the real, the rate cycle, and the October-2026 election, not by anything management decides.

The valuation is the crux. On the ADR, ITUB trades at ~10.8× earnings and ~2.15× book — the ~97th percentile of its own decade on P/E, P/B and P/S. The Gordon model, fed management’s own ~22.5% ROE and ~14.5% cost of equity, lands at ~2.2–2.45× book — i.e., the richest-ever multiple is fair, not absurd, provided a durable low-20s ROE and a falling cost of equity. The bear case needs only reversion toward the ~1.5–1.9× book ITUB itself carried in 2016–2019 (a 30–45% de-rating) if ROE normalizes toward peer levels or Selic reverses. This memo argues the business quality is not in doubt; the legitimate debate is entirely about paying full fair value at peak cycle, peak multiple, and pre-election — in a currency that can erase the local-currency compounding.

2. Business Overview

What Itaú Unibanco is. Itaú Unibanco Holding S.A. (NYSE: ITUB; B3: ITUB4) is the largest bank in Latin America and the largest private-sector bank in Brazil — a full-service universal financial conglomerate formed by the 2008 merger of Banco Itaú and Unibanco. It offers the complete banking shelf: retail and corporate deposits, credit (cards, personal, payroll, vehicle, mortgage, SME and large-corporate lending), merchant acquiring (Rede), asset management (Itaú Asset, ~R$1.24 trillion AUM), a top-tier investment bank (Itaú BBA), private banking/wealth, brokerage, insurance and private pensions (including a 30.7% stake in Porto Seguro). It operates across Brazil plus a Latin-American footprint (Chile — the merged Itaú-Corpbanca; Colombia; Argentina; Paraguay; Uruguay; Panama) and international units serving cross-border and treasury clients (US, Europe, Asia). The company is controlled through IUPAR, the vehicle jointly held by the Moreira Salles family (via Itaúsa) and the Egydio de Souza Aranha/Villela family — a durable, long-horizon controlling structure that has run the bank for generations. CEO is Milton Maluhy Filho. Headcount is ~99,600, with roughly 2,277 branches and client-service branches in Brazil plus ~16,437 ATMs [StockTitan 6-K, FY2025 Reference Form; accessed 2026-07-04].

How it makes money — the revenue architecture (FY2025, managerial basis). Operating revenues reached R$184.4 billion, +9.1% YoY, split across three engines [Itaú FY2025 earnings release; StockTitan 6-K; accessed 2026-07-04]:

The structure matters: ~two-thirds of revenue is net interest income — the spread on a R$1,490.8 billion credit portfolio (+6.0% YoY) funded by an enormous, low-cost deposit base — but a ~one-third fee-and-insurance layer (cards, acquiring, asset management, insurance, investment banking) gives Itaú a materially more diversified, capital-light income mix than a pure lender. This fee density is a genuine quality feature and a key differentiator versus the monoline digital challengers. FY2025 recurring managerial profit was R$46.8 billion, +13.1% YoY, at a 23.4% consolidated recurring ROE (24.6% in Brazil, ~13% in Latin America) [StockTitan 6-K; FY2025 MD&A; accessed 2026-07-04].

Profit segmentation (the analytically important part). Itaú reports three segments. On revenue, Retail Business is ~R$112 billion (61%), Wholesale Business ~R$63 billion (34%), and Activities with the Market + Corporation ~R$10 billion (5%). But profit is far more skewed to wholesale: the Wholesale segment earned ~R$23.0 billion of net income (+10.0% YoY) in FY2025 — on the reported segment basis, roughly half of group profit off only about a third of revenue [StockTitan 6-K; accessed 2026-07-04]. (Interpretation: the wholesale/Itaú BBA franchise — large-corporate lending, investment banking, treasury, cash management — is disproportionately profitable and capital-efficient, while retail carries the heavier cost-to-serve and credit cost. This is central to the competitive-position thesis below: Itaú’s economic center of gravity is the affluent-and-corporate end, precisely the ground the fintechs contest least.)

Client base and the digital reality. Itaú serves ~90 million active clients (company primary-account holders in the ~55–60 million range; third-party trackers put its broader “customer” count at ~100.3 million at Q4-2025, the difference being relationships versus primary account holders) [RankingsLatAm Q4-2025; accessed 2026-07-04]. The franchise is now overwhelmingly digital: over 99% of transfers and payments are executed digitally, mobile adoption exceeds 90% of the base, and 4.6 million new digital accounts were opened in 2025 — 68% of all new accounts. The bank’s iti digital wallet/bank is its explicit mass-market, lower-income counter to Nubank, and the “iyon” super-app is the rebuild of its core app; neither has disclosed unit economics suggesting it is out-competing Nubank on cost [StockTitan 6-K; accessed 2026-07-04]. Digital migration has let Itaú shrink physical capacity while lifting efficiency to a best-in-class ~39% consolidated (~35% Brazil).

Recurring vs. cyclical. The revenue base is highly recurring and granular — millions of primary banking relationships, sticky deposits, recurring card/fee income, and a diversified insurance/asset-management annuity. The variability is not churn; it is the credit cycle (provisions) and interest rates — Brazil’s Selic policy rate near ~14.25% fattens client NII and float income but pressures household debt-service and credit quality. Itaú is one of the more rate-geared large banks in the hemisphere.

Verdict. Itaú is a genuinely diversified, recurring-revenue universal bank with an unusually high-quality income mix — two-thirds spread income layered with a deep, capital-light fee-and-insurance franchise, anchored by a disproportionately profitable wholesale/investment-banking arm. It is not a simple lender and not a fragile one: ~23–25% ROE at ~39% efficiency across a R$1.49 trillion book, on a controlled, long-horizon ownership base. The business is well-understood at the unit level (deposits × spread × credit cost, plus fees), and the diversification is a real defense against the monoline fintech attack — even as the mass-retail client count is being overtaken (§4).

3. Industry Dynamics

Structure — a fat-margin oligopoly the state is deliberately disrupting. Brazilian banking has, for decades, been one of the most concentrated and profitable large banking markets on earth: five institutions — three private (Itaú, Bradesco, Santander Brasil) and two state-controlled (Banco do Brasil, Caixa Econômica Federal) — dominate. By total assets at June 2025 the ranking is Itaú 15.0%, Banco do Brasil 14.2%, Caixa 12.2%, Bradesco 10.3%, Santander 7.2% — a top-five share of ~59% [RankingsLatAm, “Brazil’s Banking Landscape 2025”; accessed 2026-07-04]. Those incumbents historically earned some of the widest net interest spreads in the world behind high-cost branch networks — the textbook fat, stable oligopoly that invites disruption.

The disruption is real, funded, and winning the customer. The challenger cohort — Nubank, Mercado Pago, Banco Inter, C6, PicPay — has attacked the mass-retail base with a branchless, near-zero-marginal-cost model. The scoreboard is now unambiguous. By customer count at Q4-2025: Caixa 158.1m, Nubank 112.0m, Bradesco 110.5m, Itaú 100.3m — and Nubank overtook Bradesco in January 2026 to become Brazil’s second-largest financial institution by customers, having started from zero in 2013 [RankingsLatAm; OpenPR “Nubank Surpasses Bradesco”; accessed 2026-07-04]. On engagement the gap is starker: in December 2025, Nu Pagamentos processed ~276.5 million monthly transactions versus Itaú’s ~143.8 million and Bradesco’s ~103.9 million — the disruptor is now running nearly double the incumbent’s throughput. The mass-retail customer front door of Brazilian banking has already changed hands.

PIX has permanently re-plumbed the payment profit pool. The central bank’s free instant-payment rail, PIX (launched 2020), in 2025 surpassed credit cards as Brazil’s leading online payment method (42% of online purchase value) and is growing ~2.5× faster than cards; monthly PIX volume reached ~R$3.4 trillion by Q1-2026 — roughly a quarter of Brazilian GDP [PaymentsCMI; PagBrasil; BCB; accessed 2026-07-04]. For payment acquirers this is destructive (the StoneCo/PagSeguro problem — free rails cannibalize interchange and float). For a deposit-and-credit bank like Itaú the effect is more mixed: PIX commoditizes the transactional/float layer and erodes some fee income, but it does not directly attack the lending or wealth pools. Layered on top, Open Finance has reached full maturity — ~60 million active consents and ~100 billion monthly API calls, four times the size of the UK’s Open Banking — which structurally lowers switching costs and is unambiguously pro-challenger.

Regulation is aggressively pro-competition and pro-consumer. The BCB has pursued PIX, Open Finance, payment-institution licensing, and interchange reform with the explicit aim of compressing incumbent rents. Fiscal policy is a swing factor: the IOF financial-transactions tax was raised in 2025, and Selic near ~14.25% is a double-edged sword — supportive of NII today but a headwind to credit quality and volume, with an easing cycle now underway. This is a regulator that has decided the old spreads were too wide and is legislating them down.

Marathon capital-cycle read. Brazilian banking is a rare case of state-forced capital-cycle disruption. On the incumbent side there is genuine supply discipline — Itaú and peers are withdrawing branch capacity (digital now 99% of transactions) and rationalizing cost — which is capital-cycle-favorable. But on the challenger side, an enormous influx of fintech capital and a regulator actively dismantling switching costs and float economics is driving a profit-pool redistribution: branches → digital, and incumbents → Nubank/Mercado Pago. The consumer-banking profit pool itself remains large, growing (credit penetration is still well below developed-market levels) and profitable — so this is not the StoneCo-style pool-destruction seen in acquiring — but the share of that pool is migrating decisively toward the lowest-cost operator.

Verdict: structurally attractive industry — but decisively better for the low-cost digital attacker than for the incumbent. The Brazilian banking pool is large, under-penetrated on credit, and still lucrative — Itaú’s ~23–25% ROE proves the pool is fat. But the structure is deteriorating for incumbents specifically: PIX has commoditized payments and float, Open Finance has demolished switching costs, and the mass-retail customer has already migrated to Nubank on count and engagement. This is a good industry in which the best incumbent can still earn excellent returns for years — but the direction of travel on the mass-retail edge is unfavorable, and the burden of proof is on the incumbent to hold the affluent, wholesale and fee pools where its advantages remain intact.

4. Competitive Position

The moat, named (Greenwald taxonomy). Itaú’s advantage is a liability-side scale-and-cost advantage — a low cost of funds from an enormous, sticky deposit base — reinforced by economies of scale and customer captivity (switching costs), and a brand/trust intangible. The mechanism is concrete and shows up in the financials: 15% of Brazilian banking assets and the system’s largest private deposit franchise fund the R$1.49 trillion credit book cheaply; primary-banking relationships (salary/payroll accounts, direct debits, card-on-file, mortgages, embedded wealth and insurance) create real switching costs; scale amortizes technology, compliance and risk infrastructure across a vast base; and the ~39% efficiency ratio (vs. Bradesco’s ~50%) plus a ~23–25% ROE are the empirical proof the moat converts to returns. Unlike a narrative moat, this one is anchored to a financial outcome that would deteriorate without it — remove the funding-cost and scale edge and the ROE collapses toward peers.

Direct comparison — Itaú is the best-run incumbent, by a wide margin.

Sources: InfoMoney/SeuDinheiro (peer ROE); RankingsLatAm (customers/transactions); NU FY2025 20-F (Nubank); accessed 2026-07-04. Against the incumbents, Itaú’s dominance is clear — it out-earns Santander Brasil by ~6 points of ROE, Bradesco by ~8, and towers over a struggling Banco do Brasil (hit in 2025 by an agricultural-credit crisis). Itaú is the quality name in the peer group; if you must own a Brazilian incumbent, this is the one.

Pressure-testing against Nubank — the moat is durable in the core but eroding at the mass-retail edge. The honest assessment, cross-read against prior analysis of Nubank, is that Itaú has already lost the mass-retail customer-count and engagement race, and cannot win it back. Nubank now has more customers (112.0m vs 100.3m), nearly double Itaú’s monthly transaction volume, and does it at a cost-to-serve (~$0.80/customer/month) and efficiency ratio (high-teens–20%) that Itaú’s branch-legacy cost base — even at a best-in-class ~39% — structurally cannot match. Nubank earns a ~30% consolidated ROE versus Itaú’s ~23–25%, and it earns it precisely in the low-income, first-account, thin-margin cohort Itaú serves least profitably. Greenwald’s warning applies directly: a scale-based moat is worth less when a lower-cost entrant can reach minimum efficient scale in a growing, contestable market — and PIX + Open Finance have handed the entrant free payment rails and demolished the switching costs that used to lock the mass customer in. Itaú’s iti is a credible defensive response, not a reconquest.

Where the moat is fully intact — and why Itaú still earns ~24%. Itaú’s advantage is migrating up-market and into complexity, exactly where the fintechs are weakest and switching costs highest:

• Affluent and high-income retail — where relationship depth, credit lines, mortgages, wealth and private banking create genuine captivity that a card-led app does not easily displace.
• Wholesale / Itaú BBA — large-corporate lending, investment banking, treasury and cash management: ~half of group profit, a relationship-and-balance-sheet business the neobanks do not contest at all.
• Fee-and-insurance density — ~24% credit-card purchase share, ~23% acquiring, pension and asset management (R$1.24tn AUM), insurance — a diversified, capital-light annuity the monolines lack.
• Mortgages — the largest private mortgage lender (+11.2% in Q1-2026), a sticky, low-loss, multi-decade relationship anchor (~39% average LTV) that also attaches deposits and cross-sell.

These are the pools where Itaú’s scale, balance sheet, brand and relationship data remain decisive, and they are why ROE is still ~23–25% despite losing the mass-retail count.

Verdict: a durable but visibly narrowing advantage — dominant among incumbents, but no longer the low-cost leader in the market it once owned. Itaú’s moat is real, multi-sourced (funding cost, scale, switching costs, fee diversification, wholesale relationships) and financially proven. But at the mass-retail edge the moat is eroding, and honestly so: Nubank has already taken the customer-count and engagement lead with a structurally lower cost base, and PIX/Open Finance have stripped away the payment-float and switching-cost defenses. The investable question is not whether Itaú has a moat — it plainly does — but whether the up-market, fee-and-wholesale moat is deep enough to defend a ~24% ROE as the low-margin base bleeds to the challengers. The verdict: durable at the top, contested-to-losing at the bottom — a franchise defending a shrinking-share, higher-quality core rather than an impregnable one.

5. Growth History and Forward Opportunities

The record: a mature franchise compounding earnings through ROE, not through the loan book. Itaú is not, and has not for years been, a volume-growth story. Its five-year record is one of moderate asset growth converted into fast earnings growth by margin expansion, mix shift toward higher-return segments, and cost discipline. FY2025 makes the pattern explicit: the total credit portfolio grew only 6.0% to R$1.49tn, yet the financial margin with clients rose 12.1% to R$121.1bn, and the recurring managerial result rose 13.1% to R$46.8bn at a 23.4% recurring ROE, up ~120bp year-on-year [4Q25 6-K press release; accessed 2026-07-04]. Earnings are growing roughly twice as fast as the balance sheet — the signature of a bank harvesting spread and mix, not chasing share. Q1-2026 continued it: recurring result R$12.3bn, +10.4% YoY, ROE 24.8%.

Where the growth is — and is not. The FY2025 segment detail shows Itaú deliberately concentrating growth in its most defensible, highest-return pools while ceding the mass-retail low end:

[4Q25 6-K press release; accessed 2026-07-04]. The individuals book — the segment most exposed to digital-bank competition — is the laggard at ~4%, while SME (+8.8% q/q) and LatAm (+12.2% q/q) carry the growth. This is a rational defense: Itaú is not trying to out-price Nubank for the low-income client; it is defending high-income retail, small-business cross-sell, corporate relationships, and its regional footprint.

Fee, insurance and cards — the capital-light engine that is decelerating. Commissions-and-fees plus insurance reached R$58.3bn, up only 6.3% in FY2025 — a meaningful slowdown versus prior years’ double-digit fee growth and a direct read on the regulatory pressure (§8): PIX cannibalizing card interchange and float; open finance lowering switching friction. Insurance is a genuine bright spot (+17% to R$11.4bn — a high-ROE, low-capital annuity) but is not large enough to re-accelerate the group.

Client and digital growth. Itaú serves ~90 million active clients with >90% of transactions executed digitally and 150m+ monthly digital logins. Its digital-native low-cost brand, iti, is the explicit counter to the fintechs, alongside the “iyon” super-app rebuild. The candid interpretation: iti is a defensive product with no disclosed evidence of winning the mass-retail land grab back from Nubank — designed to slow the leak, not reverse it.

Forward opportunities. Four, in descending order of credibility. (1) Selic-driven margin, not volume — 2026 guidance of financial-margin-with-clients growth of 5.0–9.0% on credit growth of 5.5–9.5% [4Q25 6-K guidance; accessed 2026-07-04] confirms management expects spread and mix, not the loan book, to do the work. (2) LatAm expansion — Chile (via ~39%-owned listed Itaú Corpbanca), Colombia (~99.5%-owned sub), Uruguay (Resonet to 100% in 2024) is the fastest-growing segment but structurally lower-ROE than Brazil and adds sovereign/FX risk. (3) SME cross-sell — the +8.8% q/q pool is the most attractive organic runway. (4) Digital defense (iti/iyon) — a cost story, not a growth story.

Verdict: moderate-quality growth — real, funded, and profitable, but structurally low-single-to-mid-single-digit organic, decelerating in fees, and ceding the fastest-growing customer pool to lower-cost fintechs. This is not high-quality secular growth (that is Nubank’s profile); it is a mature, dominant incumbent defending high-return segments while the mass-market frontier migrates to challengers. Earnings growth is genuine but manufactured from margin, mix, and cost — a lever powerful in a high-Selic world that fades as rates normalize. Investors should underwrite Itaú as a capital-return + ROE-defense franchise, not a grower.

6. Financial Quality

The headline caution first: the ROE is ~23–25%, not the 39–45% that data aggregators show. ROIC/aggregator ROE divides earnings by a deflated equity line (its book_val_per_sh of R$12.36 is the wrong field; the true BVPS is R$17.79 on R$196.1bn controlling equity). The authoritative franchise ROE is the managerial recurring figure — ~23.4% consolidated in FY2025, 24.6% in Brazil, 24.8% in Q1-2026 — which is itself an elite number and the real story. All KPIs below are reconciled to the FY2025 BRGAAP Management Discussion & Analysis (6-K, 2026-02-06) and the Q1-2026 press release (6-K, 2026-05-06).

Five-year KPI trend (BRGAAP managerial; R$mn unless noted)

* Brazil, 4Q25. † Brazil, best-ever first quarter. ‡ ex-FX. 2021–23 recurring net income are the reported managerial series (approx.); 2024–25 and Q1’26 are exact.

Why ROE is structurally ~23–25% (not 39%). Three reinforcing engines: (1) Brazilian bank spreads on a ~15% Selic — financial margin with clients grew +12.1% to R$121.1bn on portfolio growth, higher liability margins and mix; Itaú funds cheaply on a vast low-cost deposit base and lends at wide risk-adjusted margins (risk-adjusted client margin ~8.6%). (2) Revenue diversification — services + insurance R$58.3bn (+6.3%), insurance result +17% to R$11.4bn, a capital-light annuity that lifts ROE above a pure lender and dampens credit cyclicality. (3) The JCP tax shield (§B below) — the effective rate fell to ~8.8% in 2025, mechanically boosting net income and thus ROE.

Does the franchise improve with scale? Yes — the efficiency ratio ground from ~45% (2021) to 38.8% consolidated / ~35% Brazil (best-ever), while non-interest expenses rose only ~5% in Q1-2026 despite heavy cloud/tech spend and the book de-risked (cost of risk held at 2.6% while NPL-90 fell from 3.0% in 2023 to 1.9%). This is positive operating leverage on a book that is simultaneously improving in quality — the mix shift toward mortgage, cards and payroll (private payroll +63% YoY) is deliberately toward higher-quality origination.

Verdict — high-quality, structurally superior economics that hold with scale. Itaú earns a genuine ~23–25% recurring ROE on ~12% CET1 — roughly double a typical developed-market bank — because it combines the widest deposit franchise in Brazilian private banking, a diversified fee/insurance overlay, and disciplined underwriting through a hard-macro cycle. The one honest caveat: a chunk of the ROE premium is macro rent (high Selic) and tax structure (JCP), both of which compress if Brazilian rates normalize toward single digits — but even stripping those, the franchise’s cost and credit discipline leave it best-in-class.

6A. Quality of Earnings

IFRS vs. managerial reconciliation — clean. BRGAAP accounting net income FY2025 was R$45,671mn; recurring managerial result was R$46,830mn — a gap of only R$1,159mn (~2.5%), entirely from transparent, mostly non-cash add-backs (goodwill amortization R$727mn, software impairment R$317mn, restructuring provision R$676mn, less a tax provision R$550mn). The FY2024 bridge is similar, and the only “gain” reversed was the R$261mn XP Inc. share sale — i.e., management removes a gain, the conservative direction. Managerial earnings track underlying accounting earnings — this is not an aggressive add-back regime.

The JCP tax-shield and effective-tax collapse (32.8% 2021 → 8.8% 2025) — explained, not a red flag. Two drivers: (i) Juros sobre Capital Próprio — Brazilian law lets a company deduct an imputed “interest on own capital” against the ~34% corporate rate and pay it to shareholders (who bear a 15% withholding); as Itaú scaled JCP distributions, the deduction grew, shrinking the effective rate. (ii) The tax effect of hedging its overseas-capital investment sits in the tax line while the FX result sits pre-tax, making the pre-tax-vs-tax geography noisy year to year even though net income is FX-neutral. The managerial framework neutralizes the hedge. Critically, this shield is being legislated away: JCP deductibility is being curtailed and CSLL raised, and management guides the 2026 effective tax rate to 29.5–32.5% — a structural EPS headwind that the market must not extrapolate away.

Payout optics. Aggregators show a >100% payout in 2025 (DPS R$4.36 vs EPS R$4.05); this reflects a lumpy R$23.6bn dividends+JCP true-up declared in 4Q25, not a steady-state distribution above earnings. The true FY2025 payout is ~72% of the recurring result (~50% ordinary + an extraordinary top-up).

Provisioning — conservative. Cost of risk held flat at 2.6% while NPL-90 fell to 1.9% — the bank held/built provisions faster than delinquency deteriorated. Itaú discloses IFRS-9 stage coverage rather than a single headline NPL-90 coverage %; the total-allowance/NPL-90 ratio is not published as one number (open item; historically ~200%+). QoE verdict: high — earnings are conservative, transparent, and not diverging from cash/underlying.

6B. Insider & 6-K Sweep

The 6-K flow since 2021 is routine: quarterly earnings packages, the April 20-F, the April AGM, and a near-continuous stream of dividend + JCP notices (Itaú pays monthly base dividends plus periodic JCP true-ups). A stock-buyback program was authorized by material fact on 2026-02-05. No M&A of note in the window. The insider Form 3/4 read is thin and neutral: ~20 Form 3 initial statements around the March-2026 board reconstitution (housekeeping); ~13 code-A compensation grants at R$0 (deferred stock, including grants to controlling-family directors Setúbal, Moreira Salles, Villela Marino); modest code-S officer diversification sales (e.g., a Retail Officer sold ~183k ADR-equivalent at ~$8.82); and no code-P open-market purchases anywhere. The controlling families hold through the Itaúsa/IUPAR holdcos, which do not transact in the Form 4 flow, so open-market insider signal is structurally absent — read nothing bullish or bearish into it.

Balance-sheet strength. CET1 12.3% (12.0% Q1-2026), Tier 1 13.8%, BIS ratio ~15% (~3.6pp above the regulatory minimum incl. buffers), LCR 215%, NSFR ~125%. This is a fortress balance sheet with modest cushion above the Board’s new 12% CET1 floor.

7. Capital Allocation

This is the section that most distinguishes Itaú — and the single strongest pillar of the bull case. Where Nubank reinvests every cruzeiro at a ~30% ROE and pays nothing, Itaú has done the mature-compounder’s job with unusual discipline: it grows risk-weighted assets only in the mid-single digits at a 23%+ ROE, refuses transformative M&A, and returns its surplus capital to owners rather than chasing low-return growth. Applying the Marathon capital-cycle lens directly: management is behaving exactly as the framework prescribes a high-return incumbent should — reinvesting only where returns clear the cost of equity, and distributing the rest.

The capital-return machine, quantified. In FY2025 Itaú distributed roughly R$33.7bn in dividends and interest-on-own-capital — equal to ~72% of the R$46.8bn recurring result [WebSearch; StockTitan; accessed 2026-07-04]. The ordinary/recurring payout is ~50%; the remainder is an extraordinary distribution deliberately front-loaded into Q4-2025 (R$23.6bn distributed in Q4-2025 alone, vs R$18.0bn in Q4-2024). Management was explicit about the logic: it executed the early distribution ahead of early-2026 regulatory events that consume capital surplus, so shareholders received the excess before it was trapped. That decision drew CET1 down toward the Board’s new 12% minimum target.

The framework is now formalized. On 2026-01-26 the Board approved an updated Shareholder Remuneration Framework: a minimum 12% CET1 target, above which surplus is distributable; the statutory 25%-of-net-income minimum dividend; a monthly advance of BRL 0.015/share; a priority preferred dividend; and — importantly — authorization for share buybacks with cancellation, which permanently shrink the count and lift per-share dividends [StockTitan; WebFetch; accessed 2026-07-04]. The distribution is sized off capitalization, profitability, growth plans, M&A needs and regulatory change — a genuine excess-capital framework, not a fixed payout.

M&A: disciplined to the point of austerity. Itaú’s deal history is a catalogue of tuck-ins and minority stakes, not empire-building: Avenue (US retail brokerage) at 35% in 2022 with a path to control; Orbia (agribusiness marketplace) ~13%; Ideal (brokerage technology); Rede (payments, wholly owned); Resonet in Uruguay taken to 100% in 2024 [MarketScreener; PRNewswire; accessed 2026-07-04]. The one large regional position — Itaú Corpbanca — is a legacy holding (~39% of the listed Chilean parent) rather than a fresh bet. Critically, Itaú fully exited XP Inc.: the ~49.9% stake bought in 2017 was spun off via XPart in 2021 and the residual eliminated by 2025. Unwinding a stake in the very fintech disrupting its brokerage business — rather than doubling down — is a mark of allocation discipline, converting a strategic dead-end into distributable capital.

Incentive alignment and control. Itaú is controlled through IUPAR, jointly held by the Moreira Salles family (via Itaúsa) and the Egydio de Souza Aranha/Villela families — owner-operators whose wealth is overwhelmingly in Itaú stock, which aligns the controllers with the very dividend stream the ADR holder receives. The governance cost (developed in §9) is that the NYSE ADR is a non-voting preferred share: minority ADR holders receive the cash flows but have essentially no say. On balance the families’ multi-decade record of conservative underwriting, high payouts, and refusal of dilutive deals has served minorities well — but the structural power imbalance is real.

Verdict: intelligent, disciplined capital allocation — the strongest element of the investment case. Itaú does what the Marathon framework asks of a high-return incumbent: reinvest only the mid-single-digit RWA growth that clears its hurdle, avoid value-destructive M&A (and actively exit XP), and return the surplus. Two honest caveats: (1) the high payout is partly a symptom of limited high-return reinvestment runway (a mature bank returning capital because it cannot compound it the way Nubank can), and (2) the tax efficiency that flattered the return is now being legislated away. Neither undermines the core judgment: management has allocated capital well.

8. Changes and Headwinds — Last Two Years

The last 24 months for Itaú have been dominated less by company-specific events than by the Brazilian macro and fiscal cycle, to which the stock is a near-pure proxy (Brazil-country factor beta ~1.42 dominating all style factors).

1. The Selic super-cycle (the dominant driver). After bottoming near 10.5% in mid-2024, the Copom lifted the Selic to 15.00% by June 2025 and held it through four meetings before beginning to ease — 14.50% (Apr-2026), ~14.25% (Jun-2026) [MercoPress 2025-12-11; Bloomberg; FocusEconomics; accessed 2026-07-04]. High rates widened Itaú’s liability margins and float income (much of the 12.1% margin growth in FY2025) but pressure debt-service capacity and credit quality. The nascent easing cycle is a double-edged change — mildly supportive for credit and loan demand, mildly compressive for spread. This is the single most important swing variable for the thesis.

2. Fiscal deterioration and the IOF/tax offensive (a genuine headwind). The Lula government, chasing a zero-deficit target, sharply raised IOF on credit and FX transactions in 2025 (portions later overturned in Congress) [EY; Mayer Brown; accessed 2026-07-04]. More damaging to Itaú specifically, the reform package raised CSLL for payment institutions and curtailed the deductibility of interest-on-net-equity (JCP) — the mechanism behind Itaú’s historically low effective tax rate. The consequence is in guidance: the 2026 effective-tax-rate range is 29.5–32.5%, far above the JCP-shielded prior levels. The market should not extrapolate the old tax-advantaged EPS.

3. Credit-cycle normalization. FY2025 cost of credit rose 6.1% to R$36.6bn, but the cost-of-credit ratio held at 2.6% and consolidated 90-day NPLs were stable at 1.9% [4Q25 6-K; accessed 2026-07-04] — a benign, well-provisioned picture, but one sitting at a mid-to-late point in the consumer-credit cycle; 2026 cost-of-credit guidance (R$38.5–43.5bn) embeds continued normalization. A change to watch, not one that has broken.

4. PIX / open-finance fee compression. The regulator’s free instant-payment rail and open-finance regime continue to erode card interchange, float and switching costs — visible in Itaú’s fee-and-insurance growth decelerating to 6.3% in FY2025 (§5). A slow-burn structural headwind, not an acute shock; Itaú weathers it far better than payments-pure peers, but it caps the capital-light fee engine.

5. Capital-return framework formalized (positive). The extraordinary early distribution in Q4-2025 and the new remuneration framework (12% CET1 floor + buyback-with-cancellation) strengthen the shareholder-return thesis (§7).

6. The 2026 general election — the overhang that dwarfs the rest. Brazil holds a general election in October 2026. The stock’s Brazil-proxy character means the fiscal/political trajectory — the durability of the fiscal framework, the currency, the rate path — will drive the ADR more than any operating decision Itaú makes. A market-friendly outcome is a re-rating catalyst; a fiscally-loose outcome is a BRL-and-multiple risk. Management continuity has been a non-event by contrast: CEO Milton Maluhy Filho remains in place with the families’ backing.

Verdict: on net, a modestly weakening backdrop that the franchise is absorbing well. The high-Selic era that flattered FY2024–25 margins is turning; the tax efficiency that flattered EPS is being legislated away; fees are compressing under PIX; and a high-stakes election looms. Against that, credit quality is pristine and capital return has been formalized and front-loaded. The changes do not break the thesis — they de-flatter it: they argue against extrapolating the peak-cycle, tax-advantaged earnings that the current near-decade-high valuation implicitly capitalizes.

9. Risk Analysis (Risk Matrix)

Itaú is a high-quality bank whose equity risk is overwhelmingly Brazil-macro and currency, not idiosyncratic credit or franchise risk. For the USD ADR holder, the currency translation is the single most under-appreciated risk: a fundamentally sound result in reais can still be dead money — or a loss — in dollars.

Catastrophic / total-loss assessment. The probability of a permanent, total capital loss is low. Itaú is a systemically important, fortress-capitalized bank (Tier 1 13.8%, BIS ~15%, LCR 215%) with three decades of through-cycle profitability and the sector’s most conservative underwriting; the controlling families’ capital is co-invested alongside minorities. A wipe-out would require a Brazilian sovereign/banking-system crisis worse than the 2015–16 recession. The realistic downside is not solvency — it is valuation and currency: the ADR sits near the richest end of its own ten-year multiple range, so a de-rating toward mid-cycle multiples, compounded by a weaker real, could produce a large drawdown (the factor model shows a lifetime max drawdown of ~−69%) without ever threatening the going concern. For the USD investor the dominant risk is #1 — being right on the bank and wrong on the real.

Verdict: The idiosyncratic, business-level risk is genuinely low — this is one of the best-underwritten banks in emerging markets. The portfolio-level risk is high and almost entirely systematic: Brazil macro, the rate cycle, fiscal/electoral politics, and the currency. An investor in ITUB is underwriting Brazil first and Itaú second.

10. Valuation Discussion (Embedded Expectations)

The valuation headline is unambiguous and is the whole debate: a genuinely elite EM bank has re-rated to the top of its own decade. On the ADR (AZI, 2-Jul-2026), ITUB trades at ~10.8× trailing earnings, ~2.15× book (BVPS ~$3.77), and ~1.28× sales — and on the AZI own-history valuation index those sit at the 96.96th (P/E), 96.80th (P/B) and 97.24th (P/S) percentiles of the last ~10 years (composite 97.0). This is not a screen-cheap EM bank; it is a great EM bank priced richer than at almost any point in its listed ADR history. Everything below answers one question: what has to be true for ~2.1–2.3× book to be the right price rather than a cycle-top?

The right lens for a bank: P/B vs. ROE (Gordon), not P/E in isolation. A bank’s justified price-to-book is a function of its through-cycle ROE, its cost of equity, and its sustainable book growth:

Justified P/B = (ROE − g) / (CoE − g)

Inputs (BRL, management’s own framework): ROE — use managerial recurring ~22.5–24.8% (management refuses ROE guidance but commits to “profitability above 20% recurrently”); ignore the aggregator 39–45%. CoE — management pegs cost of capital at ~14.5% (BRL), defensible with Selic ~14.25% and long real rates ~7% + ~4% inflation; I use 13.5–16% across scenarios. g (sustainable book growth) — ROE × (1 − payout) caps out high, but nominal g must converge toward nominal GDP; I use 7–9%.

Base-case math: (0.225 − 0.09) / (0.145 − 0.09) = 0.135 / 0.055 = ~2.45×; at g = 8%, ~2.23×. The Gordon model, fed management’s own ROE and CoE, lands almost exactly on the current ~2.1–2.3× book. That is the single most important valuation finding: the richest-ever multiple is not a bubble on the fundamentals — provided you accept a durable ~22.5% ROE and a ~14.5% (and falling) cost of equity. It becomes expensive only if ROE mean-reverts toward peer levels or the Selic tailwind reverses. The multiple is fully valued, not absurd.

The USD/ADR wrinkle (the drag every dollar investor underwrites). P/B is unit-free, so the ratio is identical in BRL or USD. But realized USD returns are not currency-invariant. BRL book value compounds at ~9% locally; with the real expected to depreciate ~4–5%/yr (inflation differential; USD/BRL ~5.2 now), USD book growth is only ~4–5%. Absent any re-rating, USD total return ≈ ~5% dividend yield + ~4–5% USD book growth ≈ ~9–10% — barely covering a ~11–12% USD cost of equity. In dollars, ITUB can be exactly right in São Paulo and still mediocre in New York if the real slides.

Comparable companies

(e) = estimate. What the comp set shows. ITUB carries a deserved premium over the Brazilian incumbents — roughly 2× Bradesco’s book multiple and ~1.5× Santander Brasil’s — earned by ~600–850bps more ROE and a cost of credit roughly half the market’s (personal NPL-90 ~5.1% vs ~9.3% system; auto ~3.5% vs ~6.2%, per the Q1 call). Against Banco do Brasil it trades at ~2.5–3× the book multiple — the private-vs-state governance discount is real and durable, so BBAS is not a “cheap peer to rotate into.” On a ROE-adjusted basis (P/B ÷ ROE), ITUB at ~0.092 is cheaper per unit of return than Nubank (~0.16 reported) and roughly in line with HDFC (~0.14) — the highest-quality, lowest-multiple large-bank franchise in this set, but also the one with the least growth optionality (Nubank grows 30–40%; ITUB grows book ~9% BRL / ~4–5% USD).

Scenario analysis (illustrative; no price target)

The scenarios say multiple risk is asymmetric to the downside: the bull requires the CoE to keep falling and ROE to hold at a level no Brazilian bank has sustained through a full cycle, while the bear only requires reversion to what ITUB itself traded at (~1.5–1.9× book) as recently as 2016–2019, when ROE was ~18–20%.

Embedded expectations — what the market is underwriting. At ~2.15–2.3× book / ~10.8× earnings, the market is pricing ITUB to (1) sustain a ~22–24% managerial ROE through a normalizing credit cycle and rising statutory taxes; (2) enjoy continued cost-of-equity compression as Selic falls toward the ~12.25% Focus consensus (falling CoE lifts justified P/B even with flat ROE); (3) hold credit quality and ~5–9% loan growth despite management’s own flagged macro deterioration; and (4) defend mass-market share against Nubank. The market is underwriting management’s stated economics correctly — it is not mispricing the quality; it is pricing durability and a benign rate/FX path as near-certainties. The debate is entirely about whether peak-cycle ROE, a peak own-history multiple, and a 2026 election year are the right moment to pay full fair value. No price target.

11. Variant Perception

Consensus view. The Street is constructive-but-not-euphoric: Itaú is the best-run bank in Brazil — highest private-sector ROE (~23–25%), lowest cost of credit, record sub-39% efficiency, a fortress balance sheet (CET1 12.3%), and a shareholder-friendly capital-return record. With Selic now easing, consensus sees a rate-cut tailwind to loan demand and cost of equity, and reads ITUB as the highest-quality way to own the Brazil recovery. The valuation is acknowledged as full but defended as “you pay up for the best franchise.” The stock’s ~4× five-year run and its position near all-time highs is the consensus expressed in price.

Strongest bull case. ITUB is a durable ~22–24% ROE compounder whose returns are structurally, not cyclically, elite: a scale + primary-relationship + underwriting cost advantage producing bad-loan ratios roughly half the market’s, funded by a granular deposit base. As Selic falls, the cost of equity compresses, which mathematically lifts the justified P/B even with flat ROE, while lower rates re-accelerate loan growth and cut cost of credit — a rare simultaneous tailwind to numerator and denominator. On a ROE-adjusted basis ITUB is cheaper than Nubank and in line with HDFC, and on absolute P/E (~10.8×) it is a fraction of any DM bank of comparable quality. If the 2026 election delivers a market-friendly outcome and the real holds, the multiple re-rates toward the ~2.5–3.1× bull zone — and you were paid a mid-single-digit dividend to wait.

Strongest bear case. You are buying peak Brazil cycle × peak own-history multiple × peak political risk, in dollars. The ~22–24% ROE is a cycle-high — no Brazilian bank has sustained it through a full downturn, management itself flagged in Q1-2026 that macro conditions have worsened, NPLs are ticking up at the SME edge, and rising CSLL/JCP taxes are a structural ROE headwind. Nubank is structurally lower-cost and taking mass-market share. Revert ROE toward the ~18–20% peers earn and hold CoE at ~15–16% (if Selic reverses on fiscal/inflation slippage), and the Gordon-justified P/B falls to ~1.2–1.5× — exactly the ~1.5–1.9× ITUB traded at in 2016–2019 — a 30–45% de-rating before the BRL drag on the USD ADR. The factor model confirms the setup: a crowded, consensus-long Brazil-macro proxy at multi-year highs with a fresh negative 3-month momentum turn, into an October-2026 election.

Factor positioning (empirical). FactorsToday is emphatic that ITUB is a leveraged Brazil-macro proxy dressed as a bank: the dominant loading is Country:Brazil at beta ~1.42, Market only ~0.64, R² ~0.62 — the single largest driver of the return is Brazil itself. The related-tickers set is a wall of Brazil exposure (EWZ, BRZU, FLBR, EWZS) plus the ADR peers (BSBR, BBD). The risk-adjusted record is that of a trade that has already worked: +32.8% (y1, Sharpe 1.01), +27.7%/yr (y3), +22.4%/yr (y5) — a multi-year one-way street up. But the near-term reads the other way: m3 −23.3% annualized (~−6% raw) with a negative Sharpe, the first cooling after the April all-time high, against a lifetime max drawdown of −69%. The tape prices continuation of the Brazil rate-cut/recovery thesis; the factor evidence says the position is crowded and consensus-long into an election year.

The 3–5 assumptions that matter most:

1. ROE durability — managerial ROE holds ~22%+ through credit normalization and rising CSLL taxes, rather than reverting toward the peer 15–18%.
2. The Selic/CoE path — easing continues toward ~12% (compressing CoE, supporting the multiple), rather than stalling or reversing.
3. BRL stability — the real holds ~5.1–5.3/USD; a slide to ~6+ (as in 2024) erases the local-currency compounding for the dollar investor.
4. The 2026 election — a benign vs. fiscal-populist outcome; a Brazil-beta of ~1.42 means ITUB moves violently on the result.
5. Competitive defense vs. Nubank — ITUB holds mass-market share and pricing rather than ceding it to structurally lower-cost digital entrants.

Falsification. Bull falsified if: managerial ROE prints below ~20% for two-plus quarters while cost of credit rises, or Selic easing stalls/reverses and the real breaks toward ~6/USD — either collapses the justified P/B below the current multiple. Bear falsified if: ITUB sustains ~22%+ ROE with stable NPLs and 5–9% loan growth through the election and the CSLL step-up while Selic drifts to ~12% — in which case the richest-ever multiple is simply fair value for a best-in-class franchise and mean-reversion never comes. The variant-perception edge is not disputing that ITUB is the best bank in Brazil (it is); it is questioning whether the moment to pay full fair value is at peak-cycle ROE, peak own-history multiple, and pre-election.

12. Fact vs. Interpretation

13. Open Questions

1. NPL-90 coverage ratio. Itaú discloses IFRS-9 stage coverage, not a single headline total-allowance/NPL-90 ratio; the precise consolidated coverage % (historically ~200%+) was not published as one number in the FY2025 MD&A. Direction: robust, but confirm from the 20-F risk notes.
2. ROTE. Itaú does not disclose a separate return on tangible equity. Goodwill/intangibles are modest relative to R$196bn equity, so ROTE ≈ recurring ROE (marginally higher, ~24–26% est.) — but this is an assumption, not a reported figure.
3. iti unit economics. Itaú has not disclosed iti customer count, cost-to-serve, or profitability on a basis comparable to Nubank — so the effectiveness of the mass-retail defense cannot be independently verified.
4. Sustainable ROE ex-macro-rent. How much of the ~23–25% ROE survives a Selic normalization toward single digits and the full JCP/CSLL tax step-up? Management’s “above 20% recurrently” is a claim, not a proof.
5. LatAm ROE drag. The fastest-growing segment (LatAm, +12.2% q/q) earns a materially lower ROE (~13%) than Brazil (~24.6%); as it grows, does group ROE structurally dilute?
6. BRL forward path. The entire USD-investor thesis hinges on the real; no bank-level analysis resolves the currency question.

14. What Must Be True

For the bull (paying ~2.15× book is a good entry):

• Durable ROE: managerial recurring ROE holds ≥22% through 2026–2027 credit normalization and the CSLL/JCP tax step-up. → Falsification: two-plus consecutive quarters of recurring ROE below ~20% with rising cost of credit.
• Falling cost of equity: Selic eases toward ~12% and the real holds ≤~5.4/USD, compressing CoE and lifting justified P/B even on flat ROE. → Falsification: Selic stalls/reverses above ~14% on fiscal or inflation slippage, or USD/BRL breaks toward ~6.
• Benign election: the October-2026 outcome preserves the fiscal framework and currency. → Falsification: a fiscally-loose result that re-widens Brazil’s risk premium.

For the bear (~2.15× book is a cycle-top):

• ROE mean-reverts toward the ~18–20% Itaú earned in 2016–2019 and peers earn now, as macro rent and the tax shield fade. → Falsification: ROE sustained ≥22% through a full rate-normalization cycle.
• Multiple de-rates toward the ~1.5–1.9× book of ITUB’s own recent history (a 30–45% drawdown), compounded by BRL depreciation on the USD ADR. → Falsification: the multiple holds ≥2× through the election on proven ROE durability.
• Nubank keeps taking profitable share, not just the low-income base, forcing Itaú to defend on price and compressing the fee/retail ROE. → Falsification: Itaú’s high-income/SME/fee pools hold share and margin against the digital cohort.

The single most decisive variable common to both: whether ~22–24% ROE is structural or cyclical. Everything else — the multiple, the dividend, the USD return — is downstream of that one question, with the real as the amplifier.

15. Source Appendix

See the Source Appendix below for the full, categorized citation list (primary filings, quantitative feeds, industry/regulatory sources, and peer references). Primary sources — the FY2025 20-F, the 4Q25 and Q1-2026 6-K financial packages, and the 2026-01-26 remuneration-framework material fact — take precedence over secondary and aggregator data throughout; every material number in this article reconciles to an Itaú filing.

APPENDIX A — Standard Diligence Questionnaire

Itaú Unibanco Holding S.A. (NYSE: ITUB) · Report date 2026-07-04 · Supplemental to the research memo. Fact/Interpretation/Assumption labels applied where they matter. Where a question does not map to a bank, the correct sector analog is given.

General

What thoughtful questions have other investors asked about this company? The recurring institutional debates: (1) Is the ~23–25% ROE structural or a peak-Selic/tax-shield artifact? (2) Can Itaú defend profitable share against Nubank, or only the low-margin base it is already losing? (3) How much of the JCP tax benefit survives the 2025–26 tax reform? (4) For a USD investor, is this a bank bet or a BRL bet? (5) Is the extraordinary-dividend capacity durable or a one-off drawdown of surplus capital? This memo answers each explicitly (§6, §4, §6A/§8, §9/§10, §7).

Cyclicality & Earnings Nature

Cyclical high or low? Interpretation: high. Earnings are flattered by a ~15% Selic (wide liability spreads), benign-but-late-cycle credit (NPL-90 1.9%, a multi-year low), and a JCP tax shield (~9% effective rate) that is being legislated away. Driven by external environment or internal action? Both — internal cost discipline (efficiency ~39%→~35% Brazil) and underwriting are genuine and durable; but the level of ROE owes materially to the macro rate cycle and tax structure. How stable are revenues? Highly recurring at the franchise level (millions of primary relationships, sticky deposits, fee/insurance annuity); the variability is the credit cycle and rates, not churn. Market outlook — growing/shrinking, domestic/international? Brazilian credit is under-penetrated vs. developed markets (structural growth) but the incumbent’s share of the mass-retail pool is shrinking to fintechs; LatAm (Chile/Colombia/Uruguay) is the international growth edge, at lower ROE.

Business Quality & Competitive Moat

Industry getting more or less competitive? More — PIX, Open Finance and a well-funded fintech cohort (Nubank, Inter, Mercado Pago, C6, PicPay) are compressing incumbent rents and switching costs (§3). How profitable is the business? Elite: recurring ROE ~23–25%, ROA ~1.6% — roughly double a developed-market bank. How profitable is the industry / barriers to entry? Historically a fat-spread oligopoly (top-five ~59% of assets); barriers (scale funding, distribution, licensing, trust) remain high at the top of the market but have fallen at the mass-retail edge where digital entry is now cheap. Easily understood? Yes — deposits × spread × credit cost, plus fees. Undermined by foreign low-cost labor? No — domestic, regulated deposit franchise. Do brands matter? Yes, for trust/affluent/wholesale; less so at the price-driven mass-retail edge. Nature of competition? Incumbent-vs-incumbent on the affluent/corporate pools; incumbent-vs-fintech on cost at the mass base. Switching costs? Real for primary/affluent/mortgage/wholesale relationships; demolished for the transactional mass customer by PIX + Open Finance.

Financial Condition & Balance Sheet

Assets not fully recognized? Interpretation: the brand, the deposit franchise and the client-data/underwriting edge are unrecognized intangibles that produce the ROE premium. Off-balance-sheet liabilities? Standard banking off-B/S (guarantees, credit commitments, derivatives) disclosed in the 20-F; nothing flagged as anomalous. How conservative is the accounting? High — managerial vs. accounting net income differ by only ~2.5% (conservative add-backs; management removes gains); provisions built faster than delinquency rose. CapEx-hungry? N/A for a bank; the analog is technology + regulatory capital intensity — heavy cloud/tech investment absorbed within a falling cost ratio, and CET1 12.3% funds RWA growth internally.

Capital Allocation & Management

FCF generation and use? Bank analog = distributable capital after RWA growth and the CET1 floor. Itaú generates capital well above its mid-single-digit RWA growth and returns the surplus — FY2025 payout ~72% of recurring earnings (~50% ordinary + a front-loaded ~R$24bn extraordinary distribution). Philosophy? Formalized (2026-01-26): 12% CET1 floor, statutory minimum dividend, monthly advances, and buyback-with-cancellation authority. Significant acquisitions? No — tuck-ins/minorities only (Avenue, Orbia, Ideal, Resonet-Uruguay to 100%); exited XP Inc.; Corpbanca is a legacy Chilean holding. Buying back shares? Yes — buyback program authorized Feb-2026 with cancellation. Issuing shares to insiders? Only routine deferred-comp grants; no large dilution. Compensation/motivation? Controlled by owner-operator families (Moreira Salles + Egydio/Villela via Itaúsa/IUPAR) whose wealth is in the stock — aligned to the dividend stream; the offset is minority ADR holders own non-voting preferred (no vote). Interpretation: capital allocation is the strongest pillar of the case.

Valuation & Market Data

ADR, MLP, or K-1? ADR — 1 ADR = 1 preferred share (ITUB4), non-voting; no K-1 (issues a 1099/qualified-dividend treatment; 15% Brazilian withholding on JCP applies). Dividend policy? Monthly base dividend + periodic JCP true-ups + extraordinary distributions above the 12% CET1 floor; trailing yield ~mid-single-digit %. How profitable? ~23–25% recurring ROE (§6). Net income diverging from cash? No material divergence; managerial ≈ accounting earnings; the >100% aggregator payout is a 4Q lumpiness artifact, not earnings-quality (§6A).

Risks & Downside

What would cause the stock to decline? A BRL slide (the dominant USD-return risk), Selic reversal, a fiscally-loose 2026 election outcome, ROE mean-reversion toward peers, accelerating Nubank share loss, or a de-rating from the ~97th-percentile multiple toward mid-cycle (§9/§10). Risk of catastrophic loss? Low — systemically important, fortress-capitalized (Tier 1 13.8%, BIS ~15%, LCR 215%), three decades of through-cycle profitability. Chance of total loss? Very low — would require a Brazilian sovereign/banking crisis worse than 2015–16. The realistic downside is a large drawdown (lifetime max −69%) via valuation + currency, not insolvency.

Recent News & Events

Environment changed recently? Yes — Selic peaked at 15% (Jun-2025) and began easing (~14.25% by Jun-2026); the IOF/CSLL/JCP tax offensive raised the go-forward tax rate to 29.5–32.5%; PIX overtook cards as the leading online payment method; Nubank overtook Itaú on Brazil customer count (Jan-2026); and a general election looms in October 2026. Significant acquisitions? None material; XP fully exited. Accounting-policy change? None flagged. Recent changes — markets/facilities/management? New shareholder-remuneration framework (Jan-2026) + buyback authorization (Feb-2026); board reconstituted at the April-2026 AGM; CEO Milton Maluhy Filho continues; LatAm footprint expanded (Resonet-Uruguay to 100%, 2024).

APPENDIX B — Source Appendix

Itaú Unibanco Holding S.A. (NYSE: ITUB) · Report date 2026-07-04. Primary sources take precedence; aggregator/third-party data is reconciled to filings throughout. All URLs accessed 2026-07-04 unless noted.

1. Primary filings — SEC EDGAR (CIK 0001132597)

• Form 20-F, FY2025 (annual report, filed 2026-04-30): https://www.sec.gov/Archives/edgar/data/1132597/000113259726000132/itub-20251231.htm
• 4Q25 / FY2025 financial package (6-K, filed 2026-02-06) — press release, MD&A (BRGAAP managerial), complete financial statements, Pillar 3, institutional presentation. MD&A: https://www.sec.gov/Archives/edgar/data/1132597/000113259726000057/managementdiscussionanal.htm
• Q1-2026 financial package (6-K, filed 2026-05-06/05-11) — press release, MD&A, complete financials. Press release: https://www.sec.gov/Archives/edgar/data/1132597/000113259726000153/pressreleaseontheresults.htm
• Shareholder Remuneration Framework / stock-buyback material fact (6-K, 2026-01-26 / 2026-02-05): accession 000113259726000054.
• Prior 20-Fs (FY2024 filed 2025-04-28; FY2023; FY2022; FY2021) — multi-year trend and business/competition/regulation/risk sections.
• Form 3/4 corpus (director/officer filings, 2021–2026) — insider read (§6B).
• Full filing index: SEC_UA="..." ./scripts/edgar.sh since ITUB 2021-07-01 (5× 20-F, 622× 6-K, Forms 3/4).

2. Quantitative data feeds (reconciled to filings)

• ROIC.ai MCP — profitability/valuation/per-share/statements (BRL, IFRS): ROE, margins, EPS, multiples. Note: ROIC book_val_per_sh (R$12.36) is an erroneous line → inflates ROE to 39–45%; true BVPS R$17.79 / equity R$196.1bn / recurring ROE ~23–25% used throughout.
• AZI valuation index (own-history percentiles, 2026-07-02): P/E 10.82× (96.96th pctile), P/B 2.15× (96.80th), P/S 1.28× (97.24th), composite 97.0. ADR price $8.12, BVPS $3.772.
• AZI price CSV (download-data.php?t=ITUB) — 5-year adjusted OHLCV for the price-action event map (§10.3.1).
• FactorsToday (/api/stock-loadings, /leaderboard, /stock-info, /related-stocks): Country:Brazil beta ~1.42 (dominant), Market ~0.64, R² ~0.62; leaderboard y1 +32.8% / y3 +27.7%/yr / y5 +22.4%/yr, m3 −23.3% ann, lifetime max DD −69%; related EWZ/BRZU/FLBR/EWZS/BSBR/BBD.

3. Industry, regulatory & market-share sources

• RankingsLatAm, “Brazil’s Banking Landscape 2025” and Q4-2025 customer/transaction data — asset-share table (Itaú 15.0%, BB 14.2%, Caixa 12.2%, Bradesco 10.3%, Santander 7.2%); customer counts (Caixa 158.1m, Nubank 112.0m, Bradesco 110.5m, Itaú 100.3m); monthly transactions (Nu 276.5m, Itaú 143.8m, Bradesco 103.9m).
• OpenPR, “Nubank Surpasses Bradesco” (Jan-2026) — Nubank #2 by customers.
• PaymentsCMI / PagBrasil / Banco Central do Brasil — PIX overtaking cards (42% of online value, 2025); ~R$3.4tn monthly PIX (Q1-2026); Open Finance ~60m consents / ~100bn monthly API calls.
• MercoPress / Bloomberg / FocusEconomics — Selic path (peak 15.00% Jun-2025; ~14.25% Jun-2026); Focus consensus ~12.25%.
• EY / Mayer Brown / BDO tax alerts — 2025 IOF increases; CSLL hike for payment institutions; JCP deductibility curtailment; 2026 effective-tax-rate guidance 29.5–32.5%.
• Brazil peer ROE (InfoMoney/SeuDinheiro): Santander Brasil 17.6%, Bradesco 15.2%, Banco do Brasil impaired (2025 agri-credit stress).

4. Company / capital-allocation sources

• StockTitan — Itaú 6-K summaries (FY2025 ROE 24.4%/23.4% recurring, segment splits, distributions, remuneration framework).
• MarketScreener / PRNewswire / FinanceMagnates / Citywire — M&A history (Avenue, Orbia, Ideal, Resonet-Uruguay; XP Inc. XPart spin-off 2021 and residual exit by 2025).
• ROIC.ai company profile — segments, footprint, CEO (Milton Maluhy Filho), IUPAR/Itaúsa control, ~99,600 employees, >55m clients.

5. Peer references (public filings)

• Nu Holdings Ltd. (NYSE: NU) — FY2025 20-F and quarterly results: the key disruptor (~135m customers, ~30% ROE, ~4.7× book); Brazil profit-pool sizing; efficiency/cost-to-serve benchmarks.
• HDFC Bank (NYSE: HDB) — FY2025 annual report / 20-F: EM-bank valuation analog (~15% ROE at 2.1× book); the “elite EM bank + local-currency drag” framing.

6. FX

• USD/BRL ~5.2 (mid-2026); exchange-rates.org BRL/USD 2025 history — used for per-ADR book value and USD-return math.

Management commentary (earnings calls, guidance) is treated as hypothesis and validated against the filings and external data per the analyst framework; where the AZI news feed is empty for this ADR, the recent-events timeline is built from the 6-K flow and the Brazil macro record.