Itaú Unibanco: A 24.3% ROE on 1.9% NPLs, Fee-and-Insurance Guidance Cut to 2–5%, and 2.28 Times Book That Already Prices Durable Excess Returns

Brazil remains the economic center. It accounted for R$1.269tn, or 83%, of the June credit portfolio and generated a 25.7% recurring ROE in the second quarter, against 24.3% consolidated. Latin America outside Brazil accounted for the remaining R$253bn of credit, while exits from Argentina and now retail Colombia and Panama show a willingness to give up geographic breadth when returns or strategic scale fail to justify the capital.

A decade ago the market attached a markedly different narrative to Itaú. The bank was once treated primarily as a Brazilian macro proxy: investors bought it when credit growth, employment, the real and confidence improved, and sold it when recession or politics damaged asset quality. Itaú today is still exposed to that cycle, but the market is increasingly paying for something more durable: the ability to produce low-to-mid-20s ROE while closing branches, shrinking headcount, funding technology, holding ample liquidity and returning most earnings to shareholders. The second quarter encapsulated that story: recurring managerial earnings reached R$12.407bn, or about US$2.43bn, annualized ROE was 24.3%, the 90-day NPL ratio stayed at 1.9%, the efficiency ratio was 37.4%, and CET1 ended at 12.3%.

There is a catch: “earnings” means three different things at Itaú. Q2 recurring managerial profit was R$12.407bn; BRGAAP accounting net income was R$12.181bn. Management’s managerial statement moves items between lines, including the tax consequences of hedging investments abroad: R$1.938bn was moved from tax into financial margin in Q2. Management also excludes items such as goodwill amortization, tax and civil events and restructuring. Under IFRS, the 2025 annual figures differ again: attributable net income was R$44.857bn, versus R$46.8bn of recurring managerial result. The valuation in this report starts from recurring managerial profit because that is how Itaú manages and guides the franchise, then applies a normalization haircut for recurring “non-recurring” restructuring and cross-checks the result with IFRS equity and dividend capacity.

The central fact in the investment case is that Itaú’s present 24% ROE is supported by unusually clean credit performance despite years of very high Brazilian interest rates. The 90-day NPL ratio has been stuck at 1.9% for six consecutive quarters while mortgages and payroll lending, both comparatively secured or salary-linked, have gained weight. Stage-three gross exposures were R$55.9bn at June 30, and the accounting expected-loss allowance against stage-three assets was about R$31.0bn. Total managerial expected-loss reserves were roughly R$56.9bn. Using the disclosed 1.9% 90-day NPL ratio against the R$1.522tn broad credit portfolio gives an approximate conventional allowance-to-90-day-NPL coverage near 197%; the ratio is indicative because the reserve and loan universes are not perfectly matched.

The pressure point is beginning to move from headline NPLs toward early arrears and renegotiations. The 15–90-day delinquency ratio rose 10 basis points to 1.8%; small and middle-market Brazilian loans weakened as grace periods under government programs ended; renegotiated balances were R$36.3bn, 2.6% of the relevant book; and 1H26 gross write-offs were R$19.4bn. None is a crisis signal; together they say that the credit cycle is no longer improving effortlessly.

Cost-of-credit guidance is still credible. Itaú booked R$20.091bn in 1H26, so the full-year guidance of R$38.5–43.5bn requires only R$18.4–23.4bn in the second half. The R$41bn midpoint requires R$20.9bn, almost exactly a continuation of the second-quarter run rate. A genuine bear signal would be quarterly credit cost moving persistently above roughly R$11.5–12bn while 90-day NPLs break out of the 1.9–2.0% range.

Revenue is a more nuanced debate. Client financial margin rose 3.3% sequentially in Q2, helped by average credit balances and liability margin, while asset-management, brokerage/advisory and insurance businesses were growing fast in the half. Yet management cut full-year growth guidance for fees plus insurance results from 5–9% to 2–5%, citing capital-market volatility. Payments and collections also slipped 0.4% sequentially because gains in acquiring did not fully offset weaker current-account package revenues from companies. The first part looks cyclical; the second is a reminder that Pix and low-cost digital banking are permanently reducing the value of some traditional transactional fees.

Costs provide a more persuasive structural offset. Headcount fell from 95,714 to 90,429 year on year and the branch/client-service network from 2,738 to 2,210. First-half non-interest expenses nevertheless grew 3.9%, squarely inside guidance of 1.5–5.5%, because a bank can close branches while spending heavily on software, cloud infrastructure, cybersecurity, data and compensation. Itaú launched its generative-AI advisory experience inside the app in Q2. The relevant test is whether technology allows expense growth to stay below nominal revenue growth while service quality and underwriting remain intact, not whether AI cuts costs outright. So far, the 37.3% first-half efficiency ratio supports that case.

With Brazil eleven days from a presidential election, the macro backdrop matters more than usual. The Copom has now cut Selic for five consecutive meetings, most recently by 25 basis points to 13.75%; its September minutes emphasized weaker activity, gradual disinflation and continued fiscal uncertainty. The market expects only limited further easing before year-end, and the next monetary-policy meeting occurs after the election.

Polls in the week before this report show President Luiz Inácio Lula da Silva and Senator Flávio Bolsonaro effectively tied in plausible second-round matchups. A September 22 Quaest poll put Lula at 37% and Bolsonaro at 33% in the first round; Reuters’ compilation of Datafolha, Atlas and BTG/Nexus polling also showed a tight runoff. The first round is scheduled for October 4 and a second round, if needed, for October 25.

For Itaú, the election transmission mechanism runs through fiscal credibility, the real and the rate curve rather than simple left-versus-right labels. Reuters found investors skeptical that either campaign automatically fixes the debt trajectory. Bolsonaro’s advisers have discussed a debt-linked framework that could reduce real spending growth to zero as debt rises; economists have simultaneously argued that any next administration requires a large structural fiscal adjustment.

A credible fiscal tightening would probably strengthen the real and lower long rates, eventually compressing part of Itaú’s deposit-margin benefit but improving credit quality, capital-market activity and valuation. Fiscal slippage would do the reverse: high or rising rates can initially support the value of low-cost deposits, but eventually raise default costs, depress credit demand and force a higher equity discount rate. That asymmetry makes “higher Selic equals better bank earnings” too simple.

Capital return is also less abundant than the headline CET1 comparison suggests. Q2 materials show what ROE would have been with CET1 normalized to 11.5%: 25.1%. They do not identify 11.5% as management’s formal capital target. Itaú’s dividend policy instead says the Board considers a 13.5% minimum Tier 1 ratio when deciding distributions, and Q2 Tier 1 was 13.8%. On R$1.582tn of risk-weighted assets, the 30-basis-point gap is about R$4.7bn, versus R$12.7bn implied by mechanically reducing CET1 from 12.3% to 11.5%. The larger number is not freely distributable excess capital.

Even so, shareholders receive substantial cash: the 2025 gross payout was 72%. Itaú pays monthly interest on capital, and the 2026 gross monthly amount of R$0.018182 becomes R$0.015 net under the new 17.5% withholding. For a U.S. ADS holder, new nonresident dividends also face 10% withholding under the 2026 tax regime, subject to the still-unsettled credit mechanism described in the annual filing.

A R$49.5bn 2026 recurring-earnings estimate and a 70% payout give an illustrative gross distribution of R$3.14 per share, or 7.3% of the September 22 ITUB4 price. If approximately R$0.93 arrives as the already scheduled monthly and additional interest-on-capital payments and the remainder as dividends, a nonresident would retain roughly R$2.76 before depositary/currency expenses: a 6.4% net cash yield. The tax haircut is about 12.2% of the gross cash distribution in that illustration. A full recovery of the 10% dividend withholding through a future credit mechanism would add roughly 0.5 percentage point to the annual ADS yield; that recovery is not currently bankable.

The stock itself is sending an unusual two-class message. ITUB4 ended 2025 at R$39.23, rose enough that Itaú repurchased shares in February at R$45.38–49.65, fell to R$42.18 on June 30 and closed September 22 at about R$43.18. ITUB3 moved from R$36.35 at year-end to R$44.27 in June and R$47.28 by September 22. The common flipped from roughly a 7% discount to the preferred at year-end to about a 9.5% premium now. The company’s repurchase ledger confirms 36.555 million preferred shares were bought in February at a R$48.11 weighted average; it shows no subsequent 2026 open-market repurchase through the base date.

I do not find evidence for a fundamental change in ITUB3’s economic claim that justifies the whole relative move; market structure is the more persuasive explanation. IUPAR and Itaúsa together controlled more than 90% of common shares at year-end, leaving a small tradable float, whereas foreigners owned roughly 72% of outstanding preferred shares at June. The common has voting rights but both classes have 80% tag-along protection. Scarcity in ITUB3 and heavier foreign-flow/tax sensitivity in ITUB4 can generate a technical common-share premium without increasing an ADS holder’s economic rights. Itaú’s governance materials explicitly describe the structure as family-controlled.

The current equity debate is whether a 24% ROE is Itaú’s new normal or the upper end of a favorable credit-and-funding cycle. At US$8.43, the ADS is approximately 2.28 times June book value and roughly 9.6 times annualized 1H26 recurring managerial earnings. Those multiples are hardly a speculative bubble, but the price does require Itaú to preserve a large spread over a Brazilian cost of equity that I put near 14.5–15%. A bank earning 19–20% through a full cycle is worth much less than one capable of 22–23% sustainably.

My qualitative portrait is a mature cash cow with unusually strong compounding characteristics. Itaú has stopped being a growth story in the sense of taking ever more balance-sheet risk. Its compounding rests on better mix, underwriting, funding, digitization and capital return. That is a better business than the phrase “Brazilian cyclical bank” implies, and the current quotation already recognizes much of that improvement.

Company vertical history and financial evolution

Itaú’s history is the convergence of two Brazilian banking dynasties. One lineage traces to Casa Moreira Salles, established in Poços de Caldas in 1924 and tied to the Moreira Salles family’s commercial activities. The other began with Banco Central de Crédito in 1943, founded by Alfredo Egydio de Souza Aranha. Brazil’s mid-century urbanization, industrialization and recurring inflation created a need for institutions capable of collecting deposits, distributing credit and building national branch networks.

The predecessor banks were already listed long before the modern concept of a technology-company-style IPO. Itaú’s investor-relations history says its common and preferred shares have traded on the São Paulo exchange since 1944. There is no economically meaningful modern “IPO price and capital raised” to compare with today. The NYSE step came much later: preferred ADRs began trading on February 21, 2002, with one ADR corresponding to one preferred share.

The modern organization can be divided into four economically distinct stages.

National consolidation came first. Banco Federal de Crédito combined with Banco Itaú in 1964; by 1973 the institution had adopted the Itaú name and, according to the company history, was Brazil’s second-largest bank by deposits and first by branch count with 468 points of service. Unibanco emerged separately through mergers, reaching 330 branches across ten states by 1967. Both groups learned the same enduring skill: using acquisitions to put deposits, customers and distribution under a common risk and technology architecture.

The second stage was expansion beyond ordinary retail banking. Itaú launched what the company describes as Brazil’s first investment bank in 1966 and later opened operations in New York and Portugal. Through the 1990s and early 2000s, Brazil’s privatization and banking consolidation produced a long acquisition list: Banerj, Bemge, Banco del Buen Ayre, Banestado, BEG and others. The 2002 BBA Creditanstalt transaction created the platform that became Itaú BBA, giving the group a serious corporate and investment-banking franchise rather than leaving it dependent on retail spreads.

Banco1.net, which Itaú described as a bank without branches, was a particularly revealing 1995 move: early evidence that the organization did not regard physical distribution as sacred. That matters today because the current branch contraction is an acceleration of a long adaptation rather than a sudden defensive response to Nubank.

The third stage began with the November 2008 merger of Itaú and Unibanco. It was transformative because it united complementary customer bases, wholesale franchises, deposits, technology budgets and controlling families at the moment the global banking system was being reordered. Itaú’s own history describes the combination as creating Brazil’s largest private bank. The 2012 take-private of Redecard, subsequently branded Rede, increased the bank’s exposure to merchant acquiring; the 2014 CorpBanca transaction built a meaningful Chilean franchise; and a 2017 investment in XP acknowledged that investment distribution itself was becoming strategically important.

The fourth stage is the one shareholders own now: digital consolidation rather than territorial expansion. The 2019 acquisition of Zup was explicitly aimed at accelerating technology transformation. Itaú’s later investments in Avenue extended investment access to U.S. markets. By 2026, the bank was consolidating Avenue and placing generative-AI advice inside its app while shedding retail operations where scale and returns were less attractive.

That history explains why Itaú is difficult to attack with a single product. A digital entrant can undercut a credit card, checking-account fee or acquiring price; XP can take investment assets; a specialist payroll lender can grow faster. Itaú can move capital, customers and data across all of them.

Price history has followed Brazil’s economic regimes more closely than the company’s century-long institutional continuity would suggest. The 2015–16 Brazilian recession punished banks on credit fears, and the 2016–19 recovery rewarded falling credit costs and political optimism. The pandemic broke that trend in 2020 as banks built provisions and investors feared deep loan losses. The following recovery brought a different problem: inflation and an aggressive Selic cycle. Itaú eventually emerged with better underwriting, stronger digital adoption and a lower physical cost base. Its current ROE has surpassed the pre-pandemic norm even though the macro environment is still restrictive. The annual-results archive and current Q2 materials show that profitability, rather than balance-sheet growth alone, has been the center of management’s recent strategy.

Financially, the most useful vertical pattern is ROE rather than “revenue,” because a bank’s balance sheet is its operating plant. Itaú entered the pandemic accustomed to returns around or above 20%; returns fell sharply during the provisioning shock and subsequently rebuilt. By 2025 recurring managerial ROE was 23.4%, with R$46.8bn of recurring profit, up 13.1%; 1H26 reached 24.5%. The increase has come while the efficiency ratio has moved into the high-30s and credit quality remained unusually stable.

The 2025 accounting transition complicates any graph that treats historical earnings as perfectly homogeneous. CMN Resolution 4,966 brought an expected-credit-loss framework into Brazilian banking accounting from January 2025, making the BRGAAP credit-loss architecture closer in concept to IFRS 9. Itaú then introduced new managerial reclassifications with Q4 2025 results and consolidated Avenue in 2026. Per-share history was separately restated after 10% and 3% bonus-share issues.

Even the bonus dates illustrate why raw price downloads require care. Itaú’s Q2 per-share note says the historical series was restated for the 10% and 3% bonuses and identifies the 2025 actions in its reporting convention, while the corporate-actions ledger gives March 17, 2025 as the event date for the 10% bonus and December 23 for the 3% action. The apparent mismatch with other disclosure dates is best understood as record/event/credit-date nomenclature, not as a difference in economics. For this report, any historical per-share comparisons rely on the company-restated basis rather than unadjusted exchange closes.

The accounting reconciliation deserves more than a footnote because it can easily create a false view of margins and taxation.

Itaú’s own reconciliation shows R$1.938bn of Q2 tax effects from hedging overseas investments reclassified into financial margin. It also adds R$226m of excluded items back to BRGAAP net income to produce the recurring managerial result. Those exclusions included R$131m of goodwill amortization, R$49m of tax and civil events and R$46m of other items in Q2; the preceding quarter contained R$783m of restructuring provisions.

This recurring-versus-accounting distinction matters for valuation. I do not regard goodwill amortization as economically equivalent to a cash credit loss. I do, however, treat persistent restructuring differently. A bank that repeatedly closes branches, changes staffing and rewrites technology systems may incur recurring restructuring costs even when each individual program is temporary. Q1 2026 had R$783m of restructuring exclusions, while Q2 2025 had R$556m. My normalized valuation haircuts headline recurring managerial earnings by roughly 1–2% rather than accepting every restructuring exclusion indefinitely.

IFRS creates a third view: FY2025 IFRS pre-tax income was R$50.250bn and attributable net income R$44.857bn. The roughly 9% IFRS tax ratio is not comparable with the managerial effective-tax guidance of 29.5–32.5%, principally because managerial presentation reallocates hedge-tax effects and other items. The gap is a presentation issue with real analytical consequences, not evidence that Itaú structurally pays a 9% tax rate on ordinary earnings. The 2025 20-F is the appropriate source for statutory cross-border investor analysis; the managerial statement is the better source for short-term operating forecasts.

The balance sheet currently looks more like a source of optionality than a source of fragility. CET1 was R$194.7bn, or 12.3% of R$1.582tn RWA; Tier 1 was 13.8%; the total capital ratio was 15.4%. Average LCR was 202% and NSFR 122.1%. Client funding stood around R$1.50tn, including R$823bn of time deposits, R$128bn of demand deposits and R$172bn of savings balances.

For a bank, conventional industrial-company cash-flow tests are misleading. Customer deposits appear as financing cash flows; loan originations consume cash; loan repayments generate it. An “operating cash flow/net income” ratio can swing violently even when underlying economics do not. Maintenance capex is equally hard to isolate because a large part of what would be factory maintenance at an industrial company appears here as technology expense, software capitalization, cybersecurity spending and branch investment.

Instead of industrial free cash flow, I use distributable earnings. On a 2026 recurring-earnings estimate of R$49.5bn and a 70% payout, around R$34.7bn would be available for distribution while roughly R$14.9bn remains to finance capital growth. The resulting gross distributable-earnings yield is approximately 6.9% on the class-adjusted equity value and 7.3% for the preferred/ADS holder. The headline recurring earnings yield on ITUB4 is roughly 10.4%; the gap is capital retention, not poor cash conversion.

That distinction also explains why a growing bank with a high ROE can still compound book value while distributing most of its profits. A 23% ROE and 30% retention rate mathematically supports book-value growth near 7% before other capital movements. Itaú does not need 15% loan growth to create value; it needs to deploy retained equity at returns well above its cost.

Business model, moat, industry and competitors

Itaú earns money from four connected economic pools. The first is the spread between what customers pay for credit and what the bank pays for funding and capital. The second is transaction and service income: cards, acquiring, current accounts, asset management, brokerage and corporate advisory. The third is insurance, pensions and capitalization products. The fourth is market-making, treasury and trading, captured in the volatile “financial margin with the market.” Q2 managerial operating revenues were R$47.985bn; client financial margin alone accounted for R$32.557bn.

Loan mix matters because it is increasingly deliberate.

Mortgages grew faster than unsecured personal loans, and the mortgage book’s portfolio loan-to-value was only 39.1% at June. Q2 mortgage originations rose 63.6% year on year. Payroll lending likewise grew quickly, but the private-sector payroll portion was only about R$22.2bn of the R$81.3bn payroll book.

That mix is one reason aggregate NPLs have held up. A mortgage with 39% current LTV and a payroll loan deducted from salary have structurally different loss characteristics from revolving card exposure. The offset is duration and pricing: mortgages tie up capital for longer, while payroll growth can become dangerous if underwriting models assume that employment relationships remain stable.

Itaú does not disclose a sufficiently long, public vintage-loss series for the fast-growing private-sector payroll product to prove that its current losses are through-cycle, which leaves a genuine blind spot. Its current scale makes the risk manageable: even a hypothetical additional 200 basis points of annual credit losses on the R$22.2bn private-payroll book would cost about R$444m before tax, around R$306m after a 31% tax assumption, less than 1% of my 2026 earnings estimate. Five percentage points of incremental loss would be more painful but still only around R$0.8bn after tax. The risk becomes important if the book keeps compounding at double-digit rates without a seasoning period.

Expected-loss staging data make the credit debate more concrete. At June 30, gross staged exposures in the disclosed accounting table were approximately R$1.170tn in Stage 1, R$60.6bn in Stage 2 and R$55.9bn in Stage 3. Stage-three expected-loss allowance was about R$31.0bn. Renegotiated operations were R$36.3bn and about 52% were classified as restructured; gross 1H write-offs were R$19.4bn.

The moat begins with funding. Itaú had about R$300bn of demand and savings deposits at June, before considering other low-beta account balances. High Selic makes low-cost funding valuable because customers do not receive the full policy rate on transaction balances. At the same time, a R$1.5tn client-funding franchise reduces dependence on wholesale markets, an advantage that has survived the rise of digital banks.

A second moat is information accumulated across products. Itaú can observe salary flows, card usage, account balances, mortgage behavior, investment assets and a corporate customer’s treasury activity. The economic value is better credit selection and more precise pricing rather than “data” as a fashionable abstract asset. The 1.9% 90-day NPL rate while the overall book grows near 10% is the evidence investors should care about.

The third is distribution across customer life cycles. A mass-market account can graduate into mortgage and wealth management; an entrepreneur can move from a business account into middle-market banking and then Itaú BBA; a wealthy family can use private banking and Avenue. The ability to keep a relationship while the customer’s financial needs change is harder to replicate than any individual app feature.

The fourth moat is regulatory and capital infrastructure. Banking licenses, payments connectivity, compliance, cybersecurity, liquidity and Basel capital are expensive fixed costs. Itaú’s LCR of 202% and large capital base allow it to keep lending when smaller competitors become funding constrained.

The weaker parts of the historic moat are plainly visible. Branch convenience is worth less in a smartphone economy, basic account fees have become difficult to defend and merchant acquiring is more price-competitive. Investment distribution is no longer captive because XP and other platforms changed customer expectations. Pix has made instant low-cost account-to-account payment a public utility rather than a proprietary bank product. Those pressures explain why maintaining a 37% efficiency ratio now requires continuous technology investment.

Nubank attacks Itaú primarily from below: simple onboarding, low servicing costs and a willingness to make a mobile account the customer’s primary financial interface. Its advantage is customer experience and operating cost, while Itaú’s advantage remains product breadth, secured credit, corporate relationships and funding depth. A Nubank customer can be economically profitable without ever needing a branch, so Itaú must ensure that its legacy infrastructure adds cross-sell or underwriting value rather than just cost.

Mercado Pago attacks from a different direction. MercadoLibre begins with commerce, sellers and payment flows and then moves into credit and financial services, which makes merchant acquiring and working-capital lending more contestable. Itaú’s answer is Rede plus corporate banking, not merely a cheaper payment terminal.

XP changed investment distribution by making product choice rather than bank balance sheet the center of the relationship. Itaú’s asset-management revenues grew 4.6% sequentially in Q2 and advisory/brokerage revenues were up 25.3% in 1H, showing that the incumbent has not simply surrendered the investment client. Avenue adds a U.S. investing route for Brazilian customers.

Kaspi and SoFi are useful strategic references rather than direct peers. Kaspi shows what a finance-plus-commerce super-app can become when transaction frequency creates a strong ecosystem. SoFi shows the U.S. version of a digital bank attempting to layer lending, deposits and investing onto a software-first customer relationship. Neither faces Brazil’s exact funding, tax or rate structure, so their valuation multiples should not be imported into Itaú.

Comparison with other incumbents is more informative. Banco do Brasil’s Q2 2026 adjusted profit was R$3.9bn and its ROE only 8.3%, with 90-day delinquency at 5.61%; agribusiness delinquency had risen to 6.27%. Itaú at the same point produced 24.3% ROE and 1.9% 90-day NPLs. The gulf is large enough that Itaú deserves a substantial P/B premium.

BTG Pactual sits at the opposite end: it earned a 26.7% ROAE in Q2, with quarterly profit of R$5.1bn and R$2.7tn of assets under management/administration in the cited report. Its model is more capital-markets, wealth and corporate-credit driven, giving it higher growth and fee optionality but a different funding and retail-credit profile. Its units were around R$61.40 when its IR page was crawled.

Bradesco is the clearest warning that incumbent scale by itself is not a moat. In 2026 it launched an equity capital increase of up to R$10bn; the first subscription stage raised R$6.5bn, with controlling shareholders prepared to contribute heavily. Itaú, by contrast, is debating excess capital and buybacks, a divergence that says more about franchise economics than simple asset size.

Santander Brasil remains a credible large-bank competitor, particularly in consumer finance, payroll, cards and SME banking, but the company's 2Q26 report page does not expose uniform metrics in the version I retrieved that would let me reproduce Itaú’s managerial definitions without mixing accounting conventions. I therefore avoid inventing a pseudo-precise cost-of-risk/efficiency comparison. Santander’s official investor-relations repository remains the appropriate primary source.

The same caution applies to Banorte, Credicorp and Bancolombia. They are useful Latin American valuation references, particularly for the price investors pay for high ROE in politically volatile emerging markets, but country-specific reserve accounting, inflation, currency, deposit concentration and tax systems make a one-decimal-point “league table” more precise than the economics justify.

The verified current cross-section captures the central positioning:

Sources are Itaú Q2 materials and contemporaneous peer result reporting. Definitions differ, especially between universal retail banks and BTG.

Itaú occupies the ecological niche of the high-return universal incumbent. It has neither Nubank’s clean-sheet cost structure nor BTG’s concentrated capital-markets economics. It has a broader profit pool than either. Its vulnerability is that specialists can skim the highest-return layer of each product one at a time; its defense is being good enough in many products that the whole relationship remains economically attractive.

Governance helps explain both the stability and the discount investors should retain. Itaú’s governance page explicitly describes family control as a source of long-term strategic continuity. At December 2025, IUPAR owned 51.71% of common shares and Itaúsa another 39.21%. IUPAR itself links the Itaúsa/Setubal-Villela side of the controlling group with the Moreira Salles interests, the family line that came through Unibanco. Minority ITUB4 and ADS holders have no ordinary voting rights, although preferred shares carry 80% tag-along protection in a control transaction.

The benefit is unusually patient control, and the cost is obvious: an ADS holder cannot vote management out in the way an owner of a widely held U.S. bank theoretically can. Independent directors and an independent-related-party framework matter, but they do not change who controls the company. The governance discount should never disappear entirely.

Capital allocation has generally been rational. The group built BBA when corporate banking scale mattered, acquired Redecard to own payment economics, took control of Avenue as cross-border investing grew, and has exited retail geographies where management apparently judged returns inadequate. The February 2026 repurchase is less impressive as a valuation signal: Itaú paid an average R$48.11 for 36.555 million preferred shares, 11% above the September 22 market price. Much of the program is designed to fund employee-share delivery and cancellation, so it should not be interpreted as management declaring R$48 to be intrinsic value.

Current fundamentals, macro cycle and catalysts

Over the latest four quarters, earnings have kept improving while the room for easy positive surprises has been diminishing. FY2025 recurring managerial profit rose 13.1% to R$46.8bn and ROE reached 23.4%. Q1 2026 recurring profit was R$12.282bn, followed by R$12.407bn in Q2. First-half earnings rose 9.1% year on year to R$24.689bn and ROE rose to 24.5% from 22.8%. Q2 itself was approximately 0.5% below the market consensus Itaú compiled, which management fairly described as in line.

Operating revenues reached R$94.807bn in the half. Client financial margin was R$64.062bn, up 4.8% in management’s comparable presentation; market margin was R$1.751bn, down slightly; fees were R$22.021bn. Non-interest expenses increased 3.9%. The basic earnings equation still works: revenue growth exceeds expense growth, while credit cost remains controlled.

In the revised 2026 guidance, what did not change is more informative than what did. Credit growth remains 5.5–9.5% consolidated and 6.5–10.5% in Brazil. Client financial margin remains targeted at 5–9% growth; market financial margin remains R$2.5–5.5bn, cost of credit R$38.5–43.5bn and expense growth 1.5–5.5%. The only reduction was fees plus insurance results, cut to 2–5% from 5–9%.

That tells me management sees the fee miss as principally a revenue-mix issue rather than evidence of a deteriorating loan book. Capital-market volatility can reverse, whereas lower current-account package revenues and structural payment competition deserve more skepticism because Pix and low-cost digital providers do not disappear when markets calm.

Credit remains the most important near-term variable. NPL creation in the cited Q2 credit-quality table was R$10.657bn and write-offs R$9.369bn, the latter down about 3% quarter on quarter. Renegotiated loans rose from R$34.8bn in Q1 to R$36.3bn. Those numbers are not yet pointing to a provisioning step-change, but the direction of early arrears makes the upper half of cost-of-credit guidance more plausible than it looked six months ago.

The first-half run rate offers a clean guidance test:

A midpoint outcome needs only a modest H2 increase from the first-half run rate. The high end would require average quarterly cost around R$11.7bn, well above Q2’s R$10.14bn.

My through-cycle conclusion is that current 24% ROE is probably somewhat above sustainable normalized profitability, but not wildly so. Itaú has improved mix and efficiency enough that a return to the mid-teens is not my central case. The more realistic normalization is toward 21–23% if credit costs rise modestly and liability margins fade as Selic eventually falls.

The key sensitivities quantify that judgment. Starting from roughly R$208.5bn of equity and a full-year pre-tax earnings capacity near R$70bn, a 10-basis-point rise in credit cost on the R$1.522tn broad risk portfolio costs about R$1.52bn pre-tax and roughly R$1.05bn after a 31% tax assumption, or about 0.5 percentage point of ROE. A R$1bn move in annual market financial margin is worth approximately R$690m after tax, or 0.3 percentage point of ROE. A one-percentage-point increase in the effective tax rate on roughly R$70bn of pre-tax earnings costs around R$0.7bn, also about 0.3–0.4 point of ROE.

These are analytical estimates, not company guidance, based on Itaú’s June equity, credit book and guided tax rate.

Rates have at least four different transmission channels. A higher Selic raises the value of demand and savings deposits that reprice below the policy rate. It also raises wholesale and time-deposit funding costs, pushes borrower debt service up, suppresses long-duration credit demand and ultimately raises defaults. A lower Selic reverses those channels while generally supporting fee-generating capital-market activity.

The low-beta deposit channel alone is large enough to matter: demand plus savings deposits were roughly R$300bn at June. If only 40–60% of that balance had effective one-year exposure to a one-percentage-point change in policy rates, the gross liability-margin sensitivity before hedges, asset repricing, tax and behavioral change would be roughly R$1.2–1.8bn annually. Actual group sensitivity will be smaller or differently timed because Itaú actively manages duration and repricing.

Selic now stands at 13.75% after the September cut. The Copom said economic activity is weakening under restrictive policy and emphasized fiscal risk. The 12-month IPCA was around 4.22% in reporting around the minutes, above the 3% target, which explains why policymakers are proceeding gradually. Brazil’s government also cut its 2026 GDP growth forecast to 2.0% on September 22.

The immediate election choice is unusually close: Reuters reported Lula and Flávio Bolsonaro statistically tied in runoff polling immediately before the report date. In the September 22 session the real traded around R$5.112 per dollar while investors digested both the Copom minutes and fresh election polling.

Three fiscal-election scenarios are more useful than assigning a partisan multiple.

A credible post-election consolidation, regardless of winner, would reduce the risk premium embedded in the long end of Brazil’s curve. The real would probably benefit, Selic could fall further, and equity cost of capital could decline. For Itaú, liability margin would eventually lose some high-rate benefit; lower provisions, stronger demand for mortgages and corporate investment, better capital-markets fees and a higher P/B multiple would probably outweigh that over 2–3 years.

Continued gradual fiscal deterioration is closer to what investors currently seem to expect. Reuters’ August analysis found market participants concerned that debt would rise under either main candidate. That scenario keeps long rates elevated, slows the easing cycle and leaves Itaú with strong deposit economics but a higher cost of equity and slowly rising credit losses.

A fiscal rupture is the true downside: a sharp real depreciation and long-rate spike would increase funding volatility and defaults, hurt securities and treasury marks, suppress fee revenue and raise Itaú’s discount rate simultaneously. Itaú BBA has itself illustrated how extreme fiscal scenarios could drive materially higher rates and a much weaker currency; this risk is nonlinear.

The stock does not look priced for rupture. A preferred-share P/B around 2.28x while ROE is 24.3% says the market discounts Brazil heavily compared with a developed-market bank but still believes Itaú’s excess returns have durability. Election uncertainty is partly in the price through the discount rate; a severe debt/fiscal shock is not.

Tax policy is already a live example of political extraction from bank shareholders. The monthly 2026 interest-on-capital amount is R$0.018182 gross and R$0.015 net, which directly reflects the new 17.5% withholding. The annual filing also describes the new 10% nonresident dividend withholding. For a preferred-share base in which foreign investors owned about 72% of outstanding shares at June, these changes alter marginal demand more than they do for closely controlled common stock.

I do not expect Itaú to abandon interest on capital just because the shareholder withholding rate rose. IOC remains relevant to the issuer because Brazilian corporate deductibility can make it efficient at the company level. The ultimate mix of IOC, dividends and repurchases depends on the interaction of bank-level tax shields, shareholder withholding and capital regulation, not solely on which cash payment has the lower headline individual tax rate.

Capital is sufficient but tighter than a superficial CET1 comparison suggests. Q2 CET1 was 12.3%, Tier 1 13.8% and BIS 15.4%. The formal dividend-policy reference available on Itaú’s IR site is a 13.5% Tier 1 minimum. The Q2 “ROE at 11.5% CET1” figure is a normalization sensitivity, not a disclosed 11.5% management capital target.

At current RWA, 13.8% versus 13.5% Tier 1 represents only around R$4.7bn of static excess before management buffers, planned growth and regulatory changes. By comparison, reducing CET1 from 12.3% to 11.5% would release about R$12.7bn, but if AT1 were unchanged it would also reduce Tier 1 by roughly 80 basis points and push the ratio below the stated distribution reference. That is why I do not include R$12.7bn of “excess cash” in valuation.

Recent capital refinancing is sensible. Itaú issued R$3bn of perpetual AT1 in the first half and subsequently called older AT1; the September Tier 2 transaction adds R$5bn and roughly 30 basis points of Tier 2 ratio while R$11bn of older Tier 2 calls remove about 70 basis points. Net, the disclosed Tier 2 operations reduce the total-capital cushion by roughly 40 basis points, all else equal, but do not directly create CET1 distributable capital. Q2’s 15.4% BIS ratio began 3.8 percentage points above the regulatory minimum including buffers.

The U.S. initiative should be valued as an option, not a base-case earnings engine. Avenue gives Brazilian customers U.S. market access; Itaú took control in early 2026 and began consolidating it. The proposed U.S. national bank, which received conditional preliminary OCC approval in August according to the company’s market disclosure, could eventually give the group a domestic U.S. regulatory platform. Federal Reserve and FDIC approvals remain necessary before launch.

The strategic purpose is clearer than the near-term economics: keep wealthy Brazilian customers inside Itaú as their assets internationalize, provide cross-border corporate/private-banking services and avoid ceding the offshore relationship to specialized platforms. I assign little explicit value today because timing, capital requirements and customer acquisition costs are not sufficiently disclosed.

Latin America should also be judged by return rather than flag count. The consolidated 24.3% ROE being below Brazil’s 25.7% shows that the non-Brazil operation is currently dilutive to group return, though the company does not provide enough like-for-like allocated equity data in its headline release to calculate a clean stand-alone ex-Brazil ROE. Exiting Argentina earlier and Colombia/Panama retail now is consistent with pruning subscale businesses while retaining Chile, Uruguay, Paraguay and regional wholesale/private banking.

The immediate catalysts are now bunched more tightly than usual: the first-round election on October 4 and possible runoff October 25 come before the next major reporting cycle. The market will then see how Brazil’s rate curve responds and whether Itaú’s Q3 credit metrics break their six-quarter stability. The next Itaú results are expected in early November; the company’s IR calendar is the authoritative source, though its dynamic future-event field did not render a date in the retrieved crawl.

A practical tracking dashboard is:

The NPL and cost thresholds test whether underwriting quality is genuinely weakening. CET1 and Tier 1 tell investors whether dividends and buybacks are sustainable. Efficiency determines whether branch/headcount reduction is translating into economic leverage. The rate and election variables reveal whether the macro environment is shifting from high-rate profitability toward either benign easing or fiscal stress.

Valuation, risks and cross-synthesis

A bank with deposits should not be valued on enterprise value/EBITDA. Debt and cash are operating raw material, not financing overlays. The appropriate framework is price-to-book versus sustainable ROE and cost of equity, residual income, normalized P/E and distributable dividend capacity.

At US$8.43 and R$43.18 per preferred share, June BVPS of R$18.92 implies a 2.28x preferred-share P/B. Annualizing 1H recurring managerial profit gives approximately R$4.48 per share of earnings capacity, putting ITUB4 around 9.6x that crude run rate. The class-adjusted whole-company market cap of roughly R$499bn is around 2.39x June equity because ITUB3 trades at a premium.

Those two P/B numbers serve different questions. The 2.28x multiple is the relevant valuation for an ADS buyer because an ADS corresponds to ITUB4. The 2.39x consolidated market value is the correct figure for comparing the market’s total value of Itaú with consolidated earnings or equity.

Historical comparisons should be treated with some humility. Pandemic provisions, CMN 4,966, managerial reclassifications, Avenue consolidation and two 2025 bonus issues all contaminate a mechanically downloaded P/B or P/E series. What is clear is that 2.28x book is far from a crisis trough. It belongs to the upper half of the range one would expect for a Brazilian incumbent whose ROE has climbed into the mid-20s. I do not assign a false one-decimal historical percentile to a series whose accounting basis changed.

A steady-state residual-income identity is useful:

P/B = (ROE - g) / (Ke - g)

With a 22.5% sustainable ROE, 6.5% long-term growth and a 14.5% cost of equity, fair P/B is about 2.0x. With a 20% ROE and a 15.5% cost of equity, the multiple collapses toward the low-1s. At its current 2.28x, the stock embeds considerable confidence that Itaú’s excess ROE survives normalization.

I use 14.5% as base BRL cost of equity, 15.5% in the conservative case and 13.5% in the optimistic case. The 13.75% Selic rate is a short-term anchor rather than a long-duration risk-free rate, but it shows why assigning Itaú a U.S.-style single-digit equity discount rate would be indefensible. Management’s own roughly 14.75% cost-of-capital reference sits close to my base estimate without determining it.

My residual-income model begins with roughly R$19.6 of estimated forward book value per share and explicitly fades current ROE. The DDM cross-check uses payout ratios around 67–72%, consistent with recent distribution behavior. The result is unusually sensitive to the long-run ROE-cost-of-equity spread, the core valuation variable.

†Approximate gross USD return assuming the BRL/US$ terminal exchange rate remains near R$5.112/US$1 and including distributions. FX depreciation would lower ADS returns; a 10% cumulative BRL depreciation over three years reduces annualized USD return by roughly 3.5 percentage points. This is valuation-scenario analysis within a research framework, not investment advice.

The DDM lands in almost the same area. My base 2026 earnings estimate of roughly R$49.5bn, a 70% payout and sustainable distribution growth around 6–7% support value in the high-R$30s to low-R$40s at a mid-14% equity discount rate, or roughly US$7.5–8.2 at R$5.112/US$. Raising the required return one percentage point destroys a surprisingly large amount of value because the bank is a long-duration stream of excess returns.

Peer comparison gives a similar answer. Banco do Brasil’s 8.3% Q2 ROE and 5.61% NPL rate justify a deep discount to Itaú. BTG’s 26.7% ROAE justifies a premium where investors believe its faster-growing capital-markets and wealth franchise is durable. Itaú deserves a premium to weaker universal incumbents, but the current multiple already recognizes that superiority.

That leaves the expectation gap quite narrow. Investors do not need Itaú to post 30% ROE to justify US$8.43. They do need returns to stay around or above 22% for years, credit cost to remain controlled, and the Brazilian equity risk premium not to blow out. The next material miss is more likely to come from one of those three variables than from a single quarterly fee line.

The most fragile base-case assumption is the persistence of the excess ROE spread. My base model has returns fading from roughly 24.5% toward 22.5% against a 14.5% cost of equity. If only 70% of the excess spread survives, the five-year ROE path falls to roughly 21.5%, 21.2%, 20.8%, 20.5% and 20.1%. The residual-income valuation falls from around US$7.9 to approximately US$6.6. That 16% drop in intrinsic value comes from a change that still leaves Itaú a profitable, well-capitalized bank.

The zero-growth test is even more sobering. If recurring earnings stay flat for three years and payout remains near 70%, an ADS holder receives roughly 7.3% gross cash yield, around 6.4% under the illustrative nonresident tax mix. With Brazilian short rates still 13.75% and the sovereign opportunity set in the low-teens, that flat-earnings equity return is inadequate compensation for bank, political and FX risk. There is no margin of safety at this buy price under a three-year no-growth case.

The independent margin-of-safety verdict is therefore: not obvious. The business quality is high; the purchase price is near my reasonable-value zone rather than below a harsh downside valuation.

Five permanent-loss risks are worth developing.

The first is a Brazilian fiscal-rate-credit spiral, to which I assign medium probability but high impact. The observable indicators are a sharply weaker real, a re-steepening long-rate curve, renewed Selic hikes, NPLs above roughly 2.2% and credit cost above R$11.5–12bn quarterly. The transmission is multiplicative: defaults raise provisions while higher discount rates compress P/B. Reuters’ election reporting supports the underlying concern that debt dynamics remain difficult under either major political outcome.

The second is a slow structural loss of the most profitable retail activities to Nubank, Mercado Pago and investment platforms. Probability is medium and impact medium-to-high over five years. The early warning is client financial-margin growth falling below nominal balance growth, fee growth remaining below inflation and Itaú being forced to spend more simply to maintain customer engagement, rather than account-count headlines. The Q2 reduction in fees-plus-insurance guidance and weakness in certain package/payment lines make this a real, though not yet thesis-breaking, risk.

The third is credit normalization proving much worse than management expects. Probability is medium, impact high. Stage 2 balances, 15–90-day arrears, renegotiations and NPL creation should move before the headline 90-day ratio. The June data contain early deterioration but no break: 15–90 days was 1.8%, renegotiated exposure R$36.3bn and 90-day NPL 1.9%.

The fourth is further bank taxation. Probability is medium and impact medium. Brazil has already increased IOC withholding and introduced nonresident dividend withholding. A fiscal adjustment based disproportionately on financial-sector taxation would lower after-tax ROE and foreign-holder yield at the same time, creating both earnings and multiple pressure.

The fifth is governance risk from concentrated control. Probability of an acute abuse appears low; impact could be high if one occurred. The observable indicators are related-party transactions, capital actions that treat classes differently or a deterioration in independent-board oversight. The current 80% tag-along right reduces but does not eliminate this risk for non-voting preferred owners.

The bull case begins with a simple fact: very few large emerging-market universal banks earn 24% ROE with 1.9% 90-day NPLs and a sub-40% efficiency ratio while paying out around 70% of earnings. Itaú has done it while reducing its physical footprint and investing in technology.

The bear case begins in exactly the same place. A 2.28x book multiple only makes sense if much of that 24% return survives. The current credit cycle is mature, fee pressure is visible, foreign investors face higher taxes, and Brazil is entering a binary political event with fiscal debt dynamics unresolved.

The vertical evidence says Itaú’s durable capability is capital allocation across banking cycles. It did not become the current institution through one product invention. It repeatedly acquired franchises, integrated them, withdrew from weaker geographies, built wholesale banking, owned payments, adapted to digital channels and kept underwriting discipline. The 2008 merger was the largest single node, but the lasting capability is organizational adaptation.

The horizontal evidence says the moat is strongest where product breadth and balance sheet matter simultaneously: affluent relationships, secured household credit, middle-market companies, corporate banking and integrated wealth. It is weakest where a customer can unbundle a single digital service with almost no switching cost.

For the next year, the decisive variables are Brazilian fiscal expectations, Q3/Q4 credit migration and the pace of Selic easing. Over three years, they become sustainable ROE, technology-driven efficiency and whether fee pools stabilize after Pix/digital disruption. Over five years, the question is whether Itaú remains the customer’s financial “home” even when that customer uses Nubank, Mercado Pago, XP and foreign investment products alongside it.

The market is probably misjudging two things in opposite directions. Bears can underestimate how much structural cost and risk improvement Itaú has achieved since the pre-digital branch era; a return to 15% ROE requires a much bigger failure than ordinary rate normalization. Bulls can underestimate how demanding 2.3x book is when the local nominal cost of equity is in the mid-teens. A 20% ROE bank can be an excellent company and still produce mediocre returns if bought at a multiple that assumes 23%.

The common-share premium reinforces the second point. ITUB3’s approximately 30% rise from its 2025 year-end close to September 22 dramatically exceeds ITUB4’s roughly 10% rise, even though their basic dividend economics are closely aligned. The premium is more plausibly a scarcity/flow phenomenon than proof that the underlying bank became 20 percentage points better for common owners. An ADS holder should not extrapolate ITUB3 momentum into preferred intrinsic value.

The pre-mortem has two concrete scripts.

In the first, the election produces no credible fiscal consolidation in 2027. The real moves materially weaker, long yields rise, the central bank pauses easing and later tightens. Itaú’s 90-day NPL rate moves from 1.9% toward 2.8–3.0%, quarterly credit costs move from roughly R$10bn to R$14–15bn, normalized ROE falls toward 17–18%, and the preferred-share multiple contracts from 2.28x to 1.4–1.5x book. Even with book growth and dividends, the ADS can lose roughly half its value in dollars because earnings, valuation and FX all move against it simultaneously.

In the second, there is no macro crisis. Nubank, Mercado Pago and investment platforms simply continue taking the easiest high-return activities while Itaú retains the capital-intensive ones. By 2029 fee growth remains below inflation, technology expense prevents the efficiency ratio from improving, ROE settles near 19–20%, and investors stop valuing the bank as a structural 23% compounder. A move from 2.3x to roughly 1.5–1.6x book would again produce a large permanent loss for an investor who paid today’s price expecting present returns to persist.

The bull reasons, each traceable to the preceding evidence, are compact:

• The 24.3% Q2 ROE is backed by a 1.9% 90-day NPL ratio and 2.7% cost of credit, rather than by obvious under-provisioning in headline data.
• Brazil generated 25.7% ROE while the efficiency ratio was only 35.5%, showing that the domestic core remains exceptionally profitable.
• Client funding of about R$1.5tn and LCR of 202% give Itaú funding resilience that smaller challengers cannot easily reproduce.
• A roughly 70% payout can coexist with mid-single-digit-to-high-single-digit book growth when ROE stays above 20%.
• Management is pruning subscale geographies while building Avenue and U.S. optionality around existing affluent and corporate relationships, limiting the amount of capital I need to assume for speculative overseas growth.

The bear reasons are equally concrete:

• Early delinquency rose to 1.8% and renegotiated balances reached R$36.3bn even while 90-day NPLs stayed flat, so deterioration can still emerge with a lag.
• Fee-plus-insurance growth guidance was cut to 2–5%, and some payments/current-account revenues are facing pressures that are partly structural rather than cyclical.
• A 2.28x preferred-share P/B requires sustainable ROE well above the cost of equity for many years; the base valuation falls to about US$6.6 if only 70% of the assumed excess-return spread survives.
• Foreign preferred shareholders now absorb 17.5% withholding on IOC and 10% on new nonresident dividends before any future credit mechanism, reducing the distributable yield that supports the valuation.
• Election and fiscal risk can hit provisions, FX and the P/B multiple at the same time, while current pricing does not look like a distressed macro valuation.

My final judgment follows from the tension between franchise quality and purchase price. Itaú has proven that an old universal bank can become meaningfully more efficient without abandoning conservative balance-sheet management. Current profitability is too strong and too broad-based to call it a temporary turnaround. The more defensible long-term classification is a mature cash generator capable of compounding book value at a respectable rate while distributing most of its profits.

At US$8.43, however, the investor is paying for much of that success in advance. My base residual-income and dividend work centers around US$7.9 per ADS, with an optimistic value around US$10.5 and a conservative value around US$6.0. The current price sits inside a reasonable hold zone because future dividends can still generate an acceptable multi-year return if ROE remains above 22%, but it does not provide the discount I require before taking Brazilian fiscal, FX and bank-credit risk.

What would change the judgment upward is a price retreat without corresponding credit deterioration, or evidence that sustainable ROE deserves to be raised above 23% while the cost of equity falls. What would change it downward is a breakout in 90-day NPLs above roughly 2.2%, two quarters of credit cost above R$11.5–12bn, Tier 1 falling through the Board’s 13.5% distribution reference without an intentional capital action, or persistent fee/expense trends that push efficiency above 40%.

【Company-profile scores】

• Fundamental quality: high
• Growth: medium
• Moat: strong
• Financial soundness: strong
• Management credibility: high
• Valuation attractiveness: medium
• Risk level: medium
• Suitable investor type: value / dividend / long-term quality investors able to accept Brazilian macro and FX risk

【Investment rating】

• Rating: Hold
• One-line thesis: A 24% ROE franchise with stable 1.9% NPLs is high quality, but 2.28x book already prices durable excess returns.
• Ideal buy price: see the required standalone line below.
• Acceptable hold price: US$6.70–9.10.
• Clearly overvalued price: US$11.60 and above.
• Current-price classification: acceptable hold.
• Whether to wait for a better price: yes. A purchase around US$4.40–4.80 would provide at least a 20% margin below the conservative valuation; a less extreme US$6 area would improve expected returns materially but would not satisfy the assignment’s strict conservative-scenario margin-of-safety rule.
• Opportunity cost of waiting: approximately a 6–7% annual net/gross cash yield range plus potential book-value compounding if Itaú sustains present profitability.
• Target holding horizon: 3–5 years.
• Expected annualized return: approximately 2% conservative, 11% base and 20% optimistic over three years under broadly stable BRL/US$; FX changes can move these returns substantially.
• Max-loss risk: approximately 45–55% in a combined Brazilian fiscal, currency and credit shock in which ROE falls below 18% and P/B compresses toward 1.4–1.5x.
• Reassessment-trigger signals: 90-day NPL above 2.2%; quarterly cost of credit above R$11.5–12bn for two quarters; efficiency above 40%; Tier 1 at or below 13.5% without deliberate payout; or sustainable Brazil ROE falling below 20%.

【Ideal Buy Price】US$4.40–4.80

Basis: at least a 20% discount to the approximately US$6.0 conservative residual-income/DDM value. The range is intentionally stricter than a normal “fair-value” entry point because the assignment defines the ideal-buy signal as requiring a margin below the conservative case, not merely below base fair value.

【Valuation Range】

• current: US$8.43 (close as of 2026-09-22)
• bear (conservative · ideal buy zone): [US$4.40, US$4.80]
• base (fair · acceptable hold zone): [US$6.70, US$9.10]
• bull (optimistic · above the clearly-overvalued line): [US$11.60, US$12.50]

Research uncertainties and sources

ROE durability is the largest uncertainty. Itaú gives unusually detailed credit data, but no filing can reveal today exactly where through-cycle Brazilian losses will settle after a long period of high rates and after the new Resolution 4,966 staging regime has seasoned. The first-half data are reassuring but still contain only about eighteen months under the new BRGAAP expected-loss framework.

A second blind spot is private-sector payroll vintages. Itaú discloses the R$22.2bn balance and rapid quarterly growth but not enough product-specific cumulative loss curves to establish a through-cycle loss rate. For that reason, the valuation does not assume that recent 14.3% quarterly growth can continue indefinitely at current credit cost.

Third, a fully like-for-like 2026 incumbent-bank table is harder than it appears because Brazilian banks use different managerial adjustments, loan-book definitions and efficiency/cost-of-risk denominators. I could verify the most decision-relevant current comparisons for Banco do Brasil and BTG Pactual directly from contemporaneous result reporting, and Santander’s official results repository, but I have deliberately not populated missing one-decimal metrics for Bradesco, Santander Brasil, Banorte, Credicorp or Bancolombia from incompatible secondary definitions merely to make a table look complete.

Fourth, the September share-count update is not perfectly observable from the repurchase ledger because employee treasury-share deliveries do not necessarily appear there as market repurchases. The IR ledger shows only the February 2026 open-market acquisition of 36.555 million preferred shares through the base date. I therefore retain the June 30 outstanding count for current market-cap calculations rather than fabricating a September count.

Fifth, the U.S. national-bank option cannot be sensibly put into present value yet. Conditional OCC progress does not reveal final Federal Reserve/FDIC approval probability, launch capital, cost base, customer acquisition or earnings timing. I assign effectively zero explicit base-case value and treat approval as upside optionality.

The principal primary research base is Itaú’s Q2 2026 Management Discussion & Analysis and BRGAAP financial statements, including the managerial-accounting reconciliation, credit staging, capital, funding and guidance disclosures.

Itaú’s investor-relations pages provide the corporate history, listing/ADR structure, shareholder distributions, buybacks, governance and event calendar. The history traces the 1924/1943 roots, successive consolidation and the 2008 merger; the dividend page confirms the one-ADS-to-one-preferred relationship, monthly IOC and capital-distribution framework; the repurchase ledger records the February 2026 purchases and bonus-share history.

The company’s results center is also the source base for FY2025 Form 20-F/IFRS figures and historical result materials.

Brazil macro and election assumptions use the September Copom reporting, September 22 FX close, Reuters election polling and Reuters reporting on the two main campaigns’ fiscal positions. Selic was 13.75%; the real ended September 22 around R$5.112/US$; and the presidential race remained statistically close shortly before the October 4 first round.

The current ADS price uses the September 22 NYSE close, and Brazilian class prices use the corresponding B3 session.

The most important departure from a superficial reading of the Q2 materials is the capital target. Those materials show ROE at an 11.5% CET1 normalization. Primary IR distribution policy instead identifies a 13.5% Tier 1 level as the Board’s capital reference for distributions. I therefore do not treat 11.5% CET1 as a formal management target or count the full 80-basis-point CET1 difference as excess distributable capital.

Other tickers mentioned

• NU.US: mobile-first Brazilian challenger attacking mass-market banking, cards and consumer credit
• MELI.US: Mercado Pago parent, competing in merchant payments, wallets and credit
• XP.US: investment platform that broke incumbent banks’ captive investment-distribution model
• BBD.US: Bradesco, major Brazilian incumbent and current contrast in capital requirements
• BSBR.US: Santander Brasil, large consumer and SME banking competitor
• BBAS3.SA: Banco do Brasil, state-controlled incumbent and current credit-quality comparison
• BPAC11.SA: BTG Pactual, high-ROE Brazilian wealth, investment-banking and corporate-credit reference
• BAP.US: Credicorp, Andean universal-bank valuation reference
• GFNORTEO.MX: Banorte, Mexican incumbent reference for emerging-market banking returns
• CIB.US: Bancolombia ADR, Colombian universal-bank reference
• KSPI.US: emerging-market finance-and-commerce super-app reference
• SOFI.US: U.S. digital banking and financial-platform reference
• JPM.US: global universal-bank quality reference, not a like-for-like Brazilian peer
• BAC.US: global large-bank valuation reference
• C.US: global large-bank restructuring and valuation reference

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.