Itaú Unibanco Holding S.A. ITUB

Key data

The business

Itaú Unibanco is Brazil's largest private-sector bank, born from the 2008 Itaú-Unibanco merger and controlled by the Moreira Salles and Egydio Souza Aranha families through the Itaúsa holding company. It serves roughly 60 million clients across checking accounts, credit cards, payroll loans, mortgages and corporate lending, plus insurance, asset management and a growing digital arm. The money is made the way every bank makes it, borrow cheap from depositors, lend dearer to households and companies, and keep the spread. Itaú does it at a scale and thrift few emerging-market banks match: a 37.3% efficiency ratio, costs as a share of revenue, means it keeps far more of each real than its peers do. The moat is the deposit franchise and plain switching-cost inertia: about 90,000 employees, 2,210 branches, and an app most Brazilians treat as the default place their salary lands. What the customer holds is an Itaú card and an app. What Itaú holds is their inertia.

The numbers

In reais the bank is compounding; in dollars it keeps disappointing. Recurring managerial profit reached R$12.4 billion in the second quarter of 2026, up about 8% on the year-ago quarter, and first-half profit hit R$24.7 billion, up 9%. But the ADR has landed under consensus three quarters in a row.

The gap is currency, not operations. Every shortfall above is a few cents, and each traces to a weaker real translating the same growing reais into fewer dollars. Strip the FX and the engine runs at roughly 25% to 26% return on Brazilian equity, about 22% consolidated, numbers that put Itaú among the most profitable large banks anywhere.

From R$100 billion in 2020 to R$168 billion in 2024, about 14% a year, with 2025 not yet in this run's filings and operating revenue up another 5.9% in the first half of 2026. This is not a fast grower, it is a wide, deep, reliable one, and the compounding shows up in book value and dividends rather than headline sales.

For a bank the only number that matters more than growth is credit, and Itaú's is improving.

Loans more than 90 days overdue held at 1.9%, down from 2.3% at the end of 2025, so the credit normalization flagged a quarter ago held rather than broke. Early delinquency, 15 to 90 days, ticked up 0.1 point to 1.8% on the Latin American book while Brazil stayed flat, the seasonal drift management called out, not a turn in the cycle. Cost of credit rose 1.9% to R$10.1 billion, in line with a bigger loan book. Capital is ample: CET1 at 12.3%, down from 13.1% a year ago as the bank pushed cash back to owners, total BIS capital at 15.4%, liquidity coverage at 202%. What this memo believes that the tape does not: the three dollar shortfalls are a currency illusion laid over a bank whose earnings and credit are both getting better, and the next quarter's ADR EPS against a steadier real is what settles it.

Management

Insiders have been sellers: five sales worth $3.1 million over the last year and no purchases, the largest a $1.6 million sale by an officer in April, with plan status not disclosed in the filings, so I cannot split the pre-scheduled from the discretionary. That is a modest signal for a bank this size and reads as routine compensation selling. The capital record speaks louder. CET1 fell 80 basis points in a year because Itaú returned cash through dividends and buybacks rather than hoarding it, the efficiency ratio improved 30 basis points to 37.3%, and the controlling families, invested through Itaúsa, have run this franchise for decades on a long horizon. Detailed pay-for-performance is not in this run's filings; the behavior that is, steady payouts funded by a 22% return on equity, is the tell.

How it fails or surprises you

The Brazilian credit cycle turns. Selic is still high and consumer indebtedness elevated, and the individuals early-delinquency ratio already edged up. If unemployment rises through 2027, the 15-90 book turns first and cost of credit climbs off R$10.1 billion a quarter, squeezing the profit that justifies a top-of-range price. Watch NPL 15-90 for individuals over the next two prints.

The real stops falling (right tail). The three dollar shortfalls are translation, not trouble. A real that merely stabilizes converts the same R$24.7 billion of first-half profit into more dollars, turning shortfalls into upside with zero operational change, on a stock the market already likes. Watch BRL/USD and Q3 2026 ADR EPS against the roughly $0.22 estimate.

You paid the peak multiple for peak returns. At 2.4 times tangible book, the top of its twelve-year range and above the roughly 2x peers, Itaú is priced for its 25% Brazil return to persist. If that return mean-reverts toward the high-teens emerging-market norm while the multiple sits at its ceiling, you lose on earnings and re-rating at once. Watch Brazil recurring ROE and CET1, both already off their highs.

Closing thoughts

This is mostly priced, with one honest swing factor. At the top of its twelve-year valuation range the market has correctly recognized Itaú as the best-run large bank in Brazil and paid up, so the edge on the operating story is small and whoever sells you the stock here is not confused about the franchise. The one thing genuinely unsettled is the currency: the buyer bets the real weakness behind three dollar shortfalls reverses, the seller bets it persists, and no amount of Brazilian profit growth resolves that until the real does. The left tail is a domestic credit cycle that turns the individuals book. The right tail is a stable real re-rating dollar earnings. With credit improving and capital ample, the fatter risk is not a blowup but paying a peak multiple and watching returns or the currency give back what you paid.

The bet is still that Itaú keeps being the bank tens of millions of Brazilians default to for their paycheck, their card, and their loan, and keeps earning about a quarter back every year on the money its owners have in it. What breaks it is the one pair a top-of-range price cannot absorb together: the individuals early-delinquency ratio climbing past 2% while the real keeps sliding. Watch NPL 15-90 and BRL/USD side by side. If delinquency holds near 1.8% and the real steadies, the dollar shortfalls become upside and the premium earns its keep; if both go the wrong way, 2.4 times book is a lot to pay for a bank whose returns are normalizing.