Itaú Delivered the Quarter It Promised. The Election Raised the Hurdle It Has to Clear.

Results vs. Consensus

Itaú reports in Brazilian reais on a recurring managerial basis, which strips extraordinary items and reallocates certain lines from the statutory accounts. The bank guides on that basis and polls its consensus panel on it, and this report uses it throughout. Where a figure is in US dollars it refers to the New York ADR. Two further adjustments matter this quarter. The retail operation in Colombia left the balance sheet on 1 August, so the bank shows some growth rates "ex-Colombia". And the full management discussion document had not reached the US filing system when this note was written, so a few balance-sheet items below carry their June values and say so.

Q3 2026 Scorecard

Every attributed reais-denominated consensus puts the quarter within 0.8% of the number, and all but one within 0.3%. The bank's own thirteen-house panel expected R$12,584m. An independent press poll stood at R$12.63bn with a 23.4% return on equity. A second survey, as cited after the print, was also at R$12.63bn, while the same survey's pre-print figure of R$12.714bn implies a 0.8% shortfall. That spread of R$12.58bn to R$12.71bn is the honest range, and 12,617 sits inside it. Individual brokers' own estimates were more scattered, some about 2% to 3% higher and others in line.

The dollar rows again sit on a vendor-compiled ADR basis with a different earnings definition, share count and translation convention, which the disclosures do not let us bridge. Vendors did not even agree on the dollar revenue figure, which ranged from $8.8bn to $9.49bn for the same quarter. We report the dollar EPS miss because the market saw it, and we decline to call it a miss on Itaú's reporting basis, because on that basis the quarter was in line.

Managerial Income Statement

The table shows growth on the reported basis. The bank's headline figures of +2.5% sequentially and +6.7% year on year move the Colombian retail result into the base quarter. The difference is small and does not change any conclusion below. The more important feature of the statement is the sequential bridge. Operating revenues rose R$1,167m. Of that, R$394m came from the financial margin with the market, which is the trading and treasury line. Cost of credit rose R$276m, non-interest expenses R$401m and revenue taxes R$98m, leaving pre-tax income up R$320m. Take the market line out of both quarters and pre-tax income fell R$74m.

First Nine Months 2026 vs First Nine Months 2025

Quality of Beat

• Revenue: Operating revenues grew 5.5% year on year, and the client margin did most of it, up R$2,071m. Commissions and fees fell 0.1% year on year. Within them, advisory and brokerage fell 26.6%, current-account fees for individuals fell 16.9% and payments and collections fell 5.8%. Asset management grew 17.8% and card issuance 5.8%. The year-on-year jaws were wide: revenues up 5.5% against non-interest expenses up 1.7%. Sequentially they closed, with revenues up 2.4% and expenses up 2.4%, the latter driven by the annual 4.6% collective wage agreement. Avenue has been consolidated since the first quarter of 2026, and the release still does not size its contribution.
• Margins: Headline net interest margin with clients slipped to 9.0% consolidated and 9.9% in Brazil, each 10bp lower than in the second quarter after three flat quarters. The risk-adjusted versions slipped to 6.1% and 6.6%. Management attributes the drift to deliberate mix: mortgage, payroll and government-guaranteed small-business lending carry lower spreads. The efficiency ratio was 37.5% consolidated and 35.7% in Brazil, each marginally worse than the second quarter and markedly better than a year ago.
• Earnings: Cleaner year on year, more fragile sequentially. Pre-tax income grew 5.9% year on year and the recurring result 6.2%, close enough that the bottom line is now tracking operations. Minority interests fell 10.1% sequentially. Return on equity fell 0.7 points to 23.6% as equity grew faster than earnings. Tangible equity rose 6.7% in the quarter, and the bank distributed little.

Key KPIs

Two lines in this table carry the quarter. The return on equity fell for a second consecutive quarter, by 0.7 points in the group and 1.2 points in Brazil, while the bank raised the cost of capital it measures that return against by 0.5 points. And the risk-adjusted client margin, the metric management asks investors to judge it by, slipped 10bp along with the headline margin. Neither move is large. Both point the same way, and in August only the headline margin was drifting.

Segment Performance

Itaú does not publish conventional business-unit revenue and margin tables. It publishes the credit portfolio by product and a capital-based "business model" view that allocates average regulatory capital and a return to each activity. Both are more useful than a standard segment table, and we work through each.

Credit Portfolio by Product

The product lines sum exactly to the individuals total. Secured lending (mortgage, payroll, vehicle) reached 55.6% of the individuals book, from 55.1% in June and 53.9% a year earlier. The mortgage book, which overtook cards in the second quarter, now leads it by R$5.4bn. The consolidated book grew 1.4% reported and 2.2% excluding the Colombian sale. That 0.8-point gap is a little wider than the roughly 0.6 points we estimated in August from the disclosed R$9.7bn of net loans, and it is the step-down we told readers not to mistake for deceleration.

Business Model View: First Nine Months 2026 vs 2025

The structure we described in August holds. Insurance and services produced R$20.0bn of the R$37.3bn nine-month result, about 54%, on about 27% of average regulatory capital at a 47.9% return. Credit produced about 39% of earnings on about 62% of the capital at 15.3%. Two things moved. The insurance and services return for nine months, 47.9%, is below the 50.0% the bank showed for the first half, which can only happen if the third quarter ran below the nine-month figure. And the credit return of 15.3% now sits on top of the new 15.25% cost of capital. Itaú's single largest use of capital is now earning almost exactly what the bank says that capital costs.

Individuals: secured growth, and a forward-looking charge

Individual lending grew 2.7% sequentially and 9.6% year on year. Mortgage grew 4.4% and payroll 3.5%, with private-sector payroll up 93.7% year on year. Cards grew 1.9%, and within the higher-income Uniclass and Personnalité segments 4.3%. Vehicle loans were flat after what management called a risk adjustment comparable to the earlier card de-risking. Personal loans grew 1.3% as eligible borrowers were moved into private payroll.

"In payroll loans, we grew 3.5% in the quarter and 16.3% year-over-year. The most important thing is to note that the portfolio continues to grow in private payroll loans. We took the lead in this private payroll loan portfolio from the start, and today we are growing 93.7% year-over-year." — Milton Maluhy Filho, CEO

Asked later how much of the provision stock reflects the 2027 outlook rather than growth, management pointed to this segment: retail profitability was adjusted in the quarter, and "part of that is the forward-looking that we have done." No figure was given.

Assessment: The mix shift continues exactly as described in August. It is the right shape for a book heading into a harder year, and it is also why the headline margin keeps slipping. The more important disclosure is the one without a number. A forward-looking overlay booked in the retail business in the third quarter is prudent, but its size is the difference between a provision charge that reflects growth and one that is starting to price 2027.

Very small, small and middle market: the forecast landed

The small-business book grew 4.2% sequentially and 15.1% year on year, again led by government-sponsored facilities, up 8.0% in the quarter and 32.4% in the year. In August management guided the 90-day delinquency ratio up from 2.0% to about 2.1% as programme grace periods expired. It printed 2.1%. Short-term delinquency in the segment fell 10bp to 1.4%.

"If you go back to last quarter's call, you will remember I said that our best expectation was a 10 basis point deterioration for this quarter. So it was absolutely in line with expectations within that SMEs dynamic. The collateralized products came out of the grace period, so delinquency begins to appear, but it is a very well collateralized portfolio that has been operating at a very good profitability level." — Milton Maluhy Filho, CEO

One qualification sits in a footnote. During the quarter the bank sold loans with a low probability of recovery, without risk retention, that would otherwise have been more than 90 days overdue at September. In small business that was R$140m, worth 5bp on the ratio. Collateralised lending now makes up 58% of the segment, from 36% at the end of 2019, and 72% in very small and small companies.

Assessment: A forecast made in August was met to the decimal, and management now guides the ratio flat at 2.1% next quarter. That is a credible bank. The loan sale is disclosed, routine and small, but without it the ratio would have printed about 2.15%. The fair reading is that the forecast was met, with a little help.

Corporate: slower, and explicitly expected to get harder

Corporate lending grew 1.0% sequentially and 9.5% year on year, well below the 4.4% sequential pace of the second quarter. Management linked it to capital-market windows and stressed its quality metrics: top-ten exposure concentration down to 14% from 20% at the end of 2019, the investment-grade share up to 78% from 69%, and its share of judicial and extrajudicial reorganisations in 2025 and 2026 at 2.3% by volume. On the five restructurings with approved plans, provisions before filing were four times the eventual incurred loss.

"And the wholesale, I expect a worse scenario in 2027 than 2026. This has to do with the interest rate that is for a long period and the dynamic of our portfolio we are going to get into this year with a portfolio that is well provisioned." — Milton Maluhy Filho, CEO

Assessment: The slowdown answers the question we raised in August, when a 4.4% quarter coincided with management calling market pricing irrational. Appetite did not drift. The new information is the outlook: management expects the wholesale credit cycle to worsen in 2027 and has already lifted wholesale provisioning. Read alongside the fall in companies-side Stage 3 coverage discussed below, this is the part of the book to watch next year.

Latin America: the disposal, cleanly done

The Latin American book fell 3.5% to R$244.2bn. Excluding the Colombian retail sale, it grew 0.8% in the quarter and 11.1% in the year. The disposal lowered the region's 15-to-90-day delinquency by 14bp to 1.9%, and its 90-day ratio held at 1.1%. Latin America contributed R$2.5bn, or 6.6%, of the nine-month result.

Assessment: No surprises. The step-down was flagged, sized and delivered. The consolidated 90-day ratio rose to 2.0% while Brazil and Latin America each held flat, because the lower-delinquency Latin American book shrank as a share of the total. That is arithmetic, not credit.

Insurance and services: still the engine, now at the floor

The guided line grew 0.2% year on year in the quarter and 3.0% for nine months. Advisory and brokerage, which had grown 25.3% in the first half, fell 26.6% year on year as debt issuance slowed. Card issuance recovered, up 7.1% sequentially. The insurance result after claims fell 1.5% sequentially as retained claims rose 17.3%.

"The main message is that we revised the commissions and fees guidance for 2026 back in the previous quarter, and our best expectation today is growth within the guidance range, but more towards the low range. I will talk about 2026 guidance shortly. Our expectation is to deliver growth close to the floor of the range in fees and result from insurance." — Milton Maluhy Filho, CEO

Assessment: The segment still earns 47.9% on its capital and still produces more than half of group profit, so the pillar holds. Its growth has stalled. In August the written release blamed capital-market volatility and the call blamed economic activity and cards. This quarter both causes showed up in the numbers, in brokerage and advisory, while the card line actually recovered. Guidance at the 2.0% floor, against 3.0% for nine months, implies a fourth quarter that shrinks slightly year on year.

Trading and the capital hedge: the quarter's swing factor

The financial margin with the market reached R$1,325m, its best quarter in the five shown, net of a R$0.6bn capital-index hedge cost. Brazil contributed R$1.7bn and Latin America R$0.3bn. The nine-month market margin of R$3.08bn already sits inside the R$2.5bn to R$5.5bn full-year range, and the trading return in the business-model view recovered to 24.2% for nine months from 21.6% in the first half.

"On NII with the market, it was a very good quarter. We were able to deliver BRL 1.3 billion, already deducting the cost of the capital ratio hedge. In Brazil, we had a very strong performance of BRL 1.7 billion." — Milton Maluhy Filho, CEO

Assessment: A good quarter on the line with the least visibility, and it arrived when it was needed. It supplied more than the entire sequential gain in pre-tax income. In August we described this line as the guide with the widest range and called fourth-quarter structured operations the only upside path. The upside came a quarter early. That is welcome, and it is also the reason not to annualise the third quarter.

Guidance & Outlook

One line moved, and it moved in the bank's favour. The new tax range sits entirely below the old floor. Management called the change mechanical, the product of a higher long-term interest rate (TJLP), the balance-sheet effects that follow, and the geographic mix of the business. In the same filing the cost of capital the bank manages to rose to about 15.25% from about 14.75%.

Implied fourth quarter: Cost of credit of R$30.5bn for nine months needs about R$10.5bn in the fourth quarter to land on the R$41.0bn midpoint, against R$10.4bn in the third, so "close to the midpoint" remains the natural path. The client margin at 5.4% has crossed the floor that it trailed in August, and holding it needs only a fourth quarter in line with the year to date. Expenses at 3.1% sit in the middle of the range. The fee line is the one that guides to its edge: management expects growth "close to the floor" of 2.0%, which against 3.0% for nine months implies a fourth quarter down roughly 1% year on year. Both portfolio lines run above the top of their ranges on a twelve-month basis and were left unchanged without comment.

Street at: The thirteen-broker panel the bank polls expected R$12,584m for the quarter, about 1.4% above the second-quarter result, and got R$12,617m. The Street was modelling modest sequential growth and got slightly more. The panel had not been asked about the cost of capital or the tax range, and both moved in the same filing.

Guidance style: Itaú again revised one line and positioned another inside its range in plain language, which is the transparent habit we credited in August. Both changes this year have run the same way. In August the fee guide came down and management asked the Street to hold the bottom line by assuming the low end of the tax range. In November the tax range itself came down. Neither move is an accounting device, and the tax effect comes from a real policy rate. But in both quarters the revenue guidance softened and the tax guidance did the balancing.

Key Topics & Management Commentary

Overall Management Tone: Management was as composed and well prepared as in August, but the emphasis moved from defending the 2026 guidance to defending the balance sheet ahead of 2027. Close to a third of the prepared remarks went to provisioning models, deferred-tax composition and liquidity, which is what a bank says when it expects to be asked about losses. On the macro the language was the most cautious since we initiated. On competition it was less alarmed than in August, and the one area where management was least forthcoming was the size of the forward-looking provisions it says it has already taken.

1. The hurdle moved: cost of capital raised to about 15.25%

The guidance filing carried a sentence that matters more than any line in the table: the bank now manages its businesses against a cost of capital of about 15.25%, up from the 14.75% it disclosed in August. Management tied it to long-term interest rates, the risk premium and the institutional environment, and said plainly that it can move again either way.

"Cost of capital is a relevant variable. We have increased the cost of capital in the margin. Facing the recent changes in the market, how do we reprice the portfolio as the demands appear? This is natural with the cost of capital, and also it will impact the prices." — Milton Maluhy Filho, CEO

The arithmetic is unforgiving in two places. The consolidated spread of return over hurdle fell from about 9.6 points to about 8.4 in a single quarter: 0.7 points from a lower return and 0.5 points from a higher hurdle. And the credit business, which uses about 62% of the bank's regulatory capital, earned 15.3% over nine months. On the new hurdle it is creating almost no value at all.

Assessment: This is the most important disclosure in the print, and it was delivered in one sentence. A bank that prices loans to its cost of capital will reprice the book upward as that cost rises, which management said it will do, and that supports spreads over time. It also means the multiple an investor can pay for a 23.6% return is lower than it was in August. On a simple sustainable-growth framework the 50bp rise alone takes about a tenth of a turn off a justified price-to-book multiple.

2. The tax range itself came down, to below the old floor

In August management asked the Street to model the effective tax rate at the low end of 29.5% to 32.5% to hold the bottom line after the fee cut. In November it moved the range to 28.0% to 29.0%. On the managerial statement the realised ratio of tax to pre-tax income was 28.4% for nine months, inside the new range.

"The other adjustment, which is mechanical, is in the effective tax rate, considering the whole interest rate dynamic, the increase in the long-term interest rate, TJLP, and all the effects this produces on the balance sheet, as well the geographic mix of our businesses. We made this adjustment." — Milton Maluhy Filho, CEO

The mechanism is specific to Brazil. Interest on own capital is a deductible distribution, and the deductible amount is capped at equity multiplied by the TJLP, so a higher TJLP enlarges the deduction and lowers the effective rate. The authorities reset the TJLP each quarter, and it rose to 9.29% for the fourth quarter from 9.14% in the third. The same higher-rate environment that weighs on credit demand therefore lightens the tax line.

Assessment: This is a genuine benefit and a partial natural hedge, and management is right that it is mechanical. It is also a policy variable, attached to a deduction that has been debated before, in a country where the market reads the re-elected government as a fiscal risk. It does not compound, and it means the 2026 bottom line has now been supported by tax guidance twice in two quarters. We take the 2026 rate as given and do not extend it into 2027 until management guides it.

3. Client margin reached its floor, with help from the calendar

The line we called the most important forward watch item in August grew 2.2% sequentially (2.6% excluding Colombia) and 5.4% for nine months, clearing the 5.0% floor it trailed at the half. The bridge shows how. Average volume added about R$0.4bn and calendar days about R$0.4bn. Latin America and structured wholesale operations added about R$0.3bn. Product mix took away about R$0.4bn and spreads and the liability margin about R$0.1bn.

"In the product mix, as you saw, not only between wholesale and retail, but also within retail, growing in more collateralized products, which naturally pulls the spread to a somewhat lower level. On the other hand, we have the calendar effect in our favor this quarter, in addition to Latin America and others also in our favor, especially the wholesale structured operations, which drives this margin growth in the period." — Milton Maluhy Filho, CEO

Assessment: The commitment is met and we grade it as met. We do not grade it as operating acceleration, because two of the three positive bridge items were extra days and a volatile wholesale line, while the two spread items were negative. The underlying picture is volume growth of about 10% offset by a mix that lowers the yield, which is the trade management has chosen. It is a sound trade, and it produces client-margin growth in the mid-single digits, not the high-single-digit midpoint of the guide.

4. The trading line carried the sequential growth

As shown in the income statement, the R$394m sequential rise in the financial margin with the market exceeded the R$320m rise in pre-tax income. Brazil delivered R$1.7bn before the R$0.6bn capital hedge, its best quarter in the series shown.

"So, it was a good quarter of NII with the market for the bank." — Milton Maluhy Filho, CEO

Assessment: Good management of the treasury book is part of the franchise, and we do not discount the line to zero. But in August management explained why it gives this line a R$3bn-wide range, and in November the line supplied more than the entire quarter's sequential growth. The core operating result, revenue less credit costs and expenses excluding the market line, was flat to slightly down from the second quarter.

5. Fees: from "lower end" to "the floor"

Management said in August that fees and insurance could trend toward the low end of the original range, and then cut the range. In November it said growth would land close to the floor of the new one. Nine-month growth is 3.0% and third-quarter growth 0.2%.

"Throughout the year, I have been telling you that fee income and economic activity go hand in hand, and we have seen slower economic activity in general, a much weaker capital market, and greater volatility, which makes performance fees difficult for all asset managers. We continue to act meaningfully to deliver quality products and solutions to our clients, with ever less dependence on fee revenues." — Milton Maluhy Filho, CEO

Assessment: "Ever less dependence on fee revenues" is a notable phrase from a bank whose capital-light fee and insurance business produces more than half of its profit. We read it as describing the deliberate repricing of account packages rather than a change of strategy. The modelling point stands: for a second consecutive quarter the fee line has come in at the bottom of whatever range was in force.

6. Credit quality: delivered as forecast, with a small assist

The consolidated 90-day ratio rose 10bp to 2.0%, while Brazil (2.1%) and Latin America (1.1%) each held flat. Individuals held at 3.7%, small business rose to 2.1% as guided, and short-term delinquency fell in the consolidated and Latin American aggregates and held at 1.7% in Brazil. Cost of credit over the portfolio stayed at 2.7% for a seventh quarter, at R$10.4bn. Against a Brazilian system excluding Itaú, over-15-day delinquency for individuals was 6.7% against 13.1%, and in cards 6.6% against 14.8%.

"Clearly, 10 basis points up or down are small variations, but our best expectation today is that this 2.1% will remain flat in the next quarter. This shows not only the solidity of the portfolio, because if you go back to the 2024 indicators, you will see that we are operating at lower levels than we used to, but above all, stability and predictability." — Milton Maluhy Filho, CEO

The loan sale noted above lowered the consolidated 90-day ratio by 2bp, Brazil by 2bp, individuals by 2bp and small business by 5bp.

Assessment: The strongest pillar in the case held again, and the forecasting accuracy is itself evidence. The gap to the system is widening rather than narrowing: management showed the market's over-15-day delinquency rising while its own stays flat. The loan sale is the kind of detail that would matter if it became a habit. One quarter of it does not.

7. Coverage: a second quarter, a second companies ratio, a second mechanical explanation

In August, companies-side Stage 2 coverage fell to 19.5% from 20.8%, and management attributed it to clients migrating to Stage 3 with their provisions. We said a second consecutive decline would stop being mechanical. It did not decline: it was 19.6% in September. Companies-side Stage 3 coverage, however, fell to 49.5% from 52.9%, and total Stage 3 coverage to 52.5% from 53.8%.

"Regarding coverage, there is only one message. If you look at corporate entities, there was the exit of a client that went into bankruptcy. It was a 100% provisioned portfolio, so the provision goes out, the coverage goes out, and since what remains has a provisioning level lower than 100%, this naturally reduces the indicator, but there is no specific effect." — Milton Maluhy Filho, CEO

Assessment: The explanation is arithmetically correct, and the Stage 2 commitment we set was met. Still, companies-side Stage 3 coverage has fallen from 57.8% a year ago to 49.5%, more than 8 points in a year, with only one quarterly increase along the way. Management also says it expects a worse wholesale year. We are not calling this a problem. We are moving it from a footnote to a line we track every quarter.

8. A provisioning tutorial, before anyone asked

The prepared remarks spent unusual time showing how the bank provisions. For individuals excluding mortgage, a borrower up to 14 days late already carries a 4% provision and one 31 days late carries 58%. Under the central bank's minimum provisioning levels, in management's words, a bank "can wait 91 days to provision 50%." Stage 2 and Stage 3 balances together are 240% of loans more than 15 days overdue. Across five large corporate restructurings, provisions before filing were four times the eventual loss.

"At 31 days past due, I already have 58% provisioning on the portfolio for individuals, and it increases over time. That is why I insist so much that short-term delinquency is the best indicator for understanding the cost of credit dynamics on the bank's balance sheet, because our models respond very quickly. We provision based on expected loss rather than incurred loss." — Milton Maluhy Filho, CEO

Assessment: Persuasive, and evidently intended for the year ahead. Front-loaded provisioning means a deterioration shows up quickly in the expense line and leaves less behind it later. Investors should expect any 2027 credit stress to appear early in Itaú's cost of credit, and should not read that early appearance as a worse book than peers that recognise losses later.

9. 2027: the most cautious outlook since we initiated

Management described a 2027 of decelerating activity, a weaker job market, high household indebtedness, a weaker currency, inflation expectations revised up, and a central bank forced to hold policy tight for longer. It cited forecasts of 1.5% growth and noted that some forecasters now expect none. It expects wholesale credit to be worse than in 2026 and corporate demand to be weaker. It gave no 2027 guidance and said the budget is still being built.

"Looking at 2027, it is a challenging scenario, as I mentioned. It is challenging because of several reasons. First, the activity is decelerating, economic activity decelerating in this quarter and the next. We will see and experience challenging numbers on the growth of GDP." — Milton Maluhy Filho, CEO

Assessment: In August the caution was about the credit system. In November it is about the economy, and it comes after an election that the market priced as fiscally negative. The bank is well placed for a hard year. The investment question is what a hard year does to a return that already slipped 0.7 points this quarter, and management declined to say whether 2027 returns will be higher or lower.

10. Capital: 12.7%, a low payout, and the distribution question deferred to February

CET1 rose 40bp to 12.7%. Earnings added about 80bp, dividends and interest on own capital took about 10bp, and risk-weighted assets, including prudential and equity adjustments, took about 40bp. Tier 1 reached 14.2% with additional Tier 1 at 1.5%. Management confirmed the extraordinary-distribution discussion for the start of 2027 and, in the translated remarks, framed it against the projected excess over a 12% level. We keep the August label for continuity, although the CEO's wording differs. He said "It is not an additional dividend. It is not extraordinary." He described the payout instead as an additional distribution of capital above that level.

"From the standpoint of distribution, we will continue to practice the same policy of doing the projection, looking at the excess to 12, and what should be the additional distribution, because our objective is not to have excess specifically if we do not have an opportunity for capital allocation. [...] I am not foreseeing anything that justifies a higher retention than what we have." — Milton Maluhy Filho, CEO

Assessment: The catalyst is intact and arguably firmer: slower loan growth in 2027 means more surplus capital, and management said it sees no reason to retain more. Two qualifications apply. The bank normalises its return at an 11.5% appetite, while the translated remarks refer to an excess over 12, so the September surplus is between about 70bp and 120bp depending on the reference. And the third year of the operational-risk phase-in will consume some of it.

11. Costs: 1.7% year on year now, more pressure next year

Non-interest expenses grew 2.4% sequentially, driven by the 4.6% collective wage agreement, and 1.7% year on year. Brazil commercial and administrative expenses were flat year on year, transactional expenses fell 0.5% and technology rose 0.1%. "Other" expenses rose 39.7% on what management called a one-off marketing effect. Nine-month growth of 3.1% sits mid-range.

"That means that we still have some cost pressure. We might see more inflation for next year. That means that for our wages and the salary, we should have more pressure. We have pressures coming from one side, but we have the discipline and the way to manage that in the other side." — Milton Maluhy Filho, CEO

Management listed the 2027 pressures: new buildings leased to adapt the hybrid-work model, continued technology and AI spending, and higher wage inflation. The finance chief contrasted cost growth of about 10% year on year in early 2025 with 1.5% now.

Assessment: The 2026 record is excellent and we would underwrite it again. The 2027 message is that the cost line will grow faster just as revenue growth slows, and management said the remaining room is in specific businesses rather than across the bank. The efficiency pillar is intact, and its contribution to earnings growth is likely to narrow next year.

12. NeoSpace and the AI claims: the first quantified model benefit

The bank acquired NeoSpace, a developer of generative-AI foundation models it had worked with for three years. Management offered its most specific AI claim to date: a 10-point improvement in the Gini coefficient of its credit models. It also reported a 50% reduction in unit cost per token since early 2025 and a target of full modernisation and mainframe retirement by 2028.

"What is the practical benefit of this acquisition? The results that we have seen in our models, it is completely different. We increased in 10 points the Gini of the models with high penetration, high accuracy." — Milton Maluhy Filho, CEO

Assessment: A 10-point Gini improvement in credit scoring is a real and testable claim, and better than the capability percentages offered in August. It is still a model statistic rather than a reported cost or revenue figure, and the acquisition adds to the technology spend that management has said will rise. We move closer to crediting the AI programme but do not yet see it in a reported line.

13. Why a bank with an in-line quarter spent the call on its balance sheet

The deck added slides on tangible equity (R$138.7bn, 60.6% of total equity), deferred tax credits (R$71.5bn net, 73% from expected-loss provisions and 2% from tax-loss carryforwards that management expects to be about zero by year end), labour and civil provision coverage, liquidity (LCR 205.8%, NSFR 123.7%) and a loan-to-funding ratio of 78.6%.

"It is very important to understand not only the P&L dynamics, but also to understand the structure of the balance sheet to show how prepared we are and how protected and reinforced the bank's balance sheet is." — Milton Maluhy Filho, CEO

Assessment: The composition of the deferred tax asset matters in a downturn. A bank whose deferred tax sits mostly in expected-loss provisions, with almost none from tax-loss carryforwards, is far less exposed to capital deductions if earnings weaken than a peer carrying large carryforwards. Management did not name peers, but the point was clearly comparative. It is also a signal of what management expects the next year to test.

Analyst Q&A Highlights

Whether 2027 is a hard year, and what the bank does about it

The first question of the call set the agenda for the rest of it. It laid out a macro case of weak growth, inflation near 6% and a policy rate in the 13% to 14% range, and asked how the bank would navigate it. Management agreed with the premise without hedging, named the transmission channels, and stated a specific view on wholesale credit.

Q: "Well, you are looking at the end of your speech with the challenging scenario that Itaú is ready. I am looking at your macro scenario. GDP half, inflation six, the Selic 13, 14%. It seems like a challenging scenario. Can you tell us, I know that this is not guidance, but what is the long road dynamics?"
— Thiago Batista, UBS

A: "There are people mentioning zero growth next year. Very early to mention a number, but more important than the number is the direction that we are headed. That produces challenges. It happens with the job market, an increase in the decrease of jobs, the indebtedness of the family is still high, the interest rates are very restrictive because of the volatility, exchange rate devalued, inflation being reviewed for 2027 at a threshold that is higher than what we expected [...]"
— Milton Maluhy Filho, CEO

Assessment: The bank's own macro view is now close to the bear case on Brazil. That is honest and it is useful. It also means the 2027 guidance in February will be built on that view, and the Street should expect growth guidance below this year's ranges.

Holding the spread over a rising cost of capital

The second question went straight to the arithmetic of Key Topic 1: slower loan growth, higher credit costs and a higher cost of capital all squeeze the gap between return and hurdle. Management named repricing and efficiency as the levers and declined to quantify either.

Q: "So, the scenario of deceleration that you described and the growing of the credit portfolio, higher cost of capital, deterioration of the cost of credit, what are the levers, well, excluding the control of cost that you see, how do you keep the spread between the profitability of the bank ROE and the cost of capital that is increasing?"
— Renato Meloni, Bernstein Autonomous

A: "The curves of the interest rates on the long-term will impact the price as the cost of funding changes. So, our portfolio is highly repricable through time. That will continue. The expansions continue to take place in the businesses that we have a high conviction."
— Milton Maluhy Filho, CEO

Assessment: Repricing is the right lever and the book does reprice. It reprices with a lag, however, and new lending will be smaller in a year of weaker demand. The answer implies a spread that narrows before it recovers.

Where the cost of capital goes from here

A question placed the new hurdle in its history, from a low near 12.9% to a high near 17%, and observed that long-term rates above 15% put the bank close to the worse end of that range. Management confirmed the new figure, described what would bring it down, and turned the point into one about the country rather than the bank.

Q: "You have in your institutional presentation a very nice graph of the records showing the ROE, the cost of capital, the generation of value, and we have seen that at the worst moment, you have a cost of capital of 17%, at the best moment, 13%. To be more exact, 12.9%. And now the long interest rates are above 15%. Probably you are going back to the next of the worst moment of the series."
— Eduardo Rosman, BTG Pactual

A: "[...] we are now operating with 15.25% of the cost of capital. That is the cost of capital now. Can it change up or down? It depends on the economic conditions, it depends on the long curve, it depends on the risk premium, it depends on the institution environment, it depends on the legal securities, all the elements that are necessary for you to work with a lower cost of capital."
— Milton Maluhy Filho, CEO

Assessment: The hurdle is set by the sovereign, not the bank, and every factor listed is outside management's control. A Brazil equity investor is being told plainly that the discount rate on this franchise now follows the fiscal path.

Whether the election changed the outlook, or only confirmed it

A question about whether 2027 would be specifically difficult, and whether the central bank might tighten capital or provisioning rules, drew the call's only direct reference to the election. Management argued that the hard year was already locked in before the vote.

Q: "My question is, do you expect a year specifically difficult for next year? Given that the other players have a strong balance, do you believe that there will be any Central Bank measure, whether if it is capital provision or any type of restriction for any type of cap?"
— Antonio, Bank of America

A: "For next year, many of the measures that will have an inflationary effect and the deceleration of the activity were already subcontracted. Regardless of the results of the elections, the scenario for 2027 was a challenging scenario, without a shadow of a doubt. Naturally, now the macroeconomists and the economists start to adjust their models and their projections. We do not work with a conservative scenario. We work with a realist scenario."
— Milton Maluhy Filho, CEO

Assessment: The answer separates the bank's operating outlook from the market's repricing, and it is credible: the cautious language on credit predates the vote. It does not address the channel through which the election hit the equity, which is the cost of capital and the currency rather than next year's loan book.

Small-business delinquency beyond the government guarantees

A question on the small-business book asserted that short-term delinquency had risen in the quarter and asked how the unguaranteed part was behaving with rates higher for longer. Management corrected the premise and restated its forecast.

Q: "Portfolio is still growing well, delinquency normal, but the delinquency on the short-term has increased on the quarter. How do you see this indicator from now on with the interest rates higher, longer, and how the part of the portfolio that doesn't have a guarantee of the government is behaving?"
— Bernardo Guttmann, XP

A: "When you look at the short-term delay, there is a drop of the 10 basis points. It's something that shows for the long-term delay, and it gives me security for doing the segment that I've done in the past."
— Milton Maluhy Filho, CEO

Assessment: The correction was accurate: small-business 15-to-90-day delinquency fell to 1.4% from 1.5%. The part of the question that went unanswered was the behaviour of the unguaranteed book, which management discussed only in general terms of rating migration and collateral.

How much of the provision stock is already pricing 2027

The sharpest credit question of the call observed that the observed indicators are stable while expected-loss provisions run near R$11bn, and asked how much of that reflects growth and how much reflects a forward-looking view. Management confirmed a forward-looking adjustment in retail without sizing it.

Q: "The overview that suggests in your forward-looking, you are capturing risks that did not appear in the observed indicators. Of course, you don't have to give me a number, specific number, but how much of the expected credit loss that you've given us is of growth, and how much is looking at 2027 for the models that you're capturing for the future problems?"
— Gustavo Schroden, Citi

A: "If you look at the model of business profitability, you will see that in the retail this quarter, we had an adjustment in the profitability. Part of that is the forward-looking that we have done. We have recorded the scenario. [...] If you ask me, "Milton, do you have any discomfort in regards to the provisions?" No, I have a comfort, excess of comfort, and that gives you degrees of freedom for you to continue to work and correct proactively."
— Milton Maluhy Filho, CEO

Assessment: "Excess of comfort" is a strong phrase from a disciplined bank, and it suggests the forward-looking charge is meaningful. It also means part of the third-quarter cost of credit was a scenario provision rather than observed loss, which flatters the forward picture if the scenario is mild and does not if it is severe.

Market share, irrationality and the cost line in a downturn

Two related questions asked whether a strong bank should take share while weaker competitors retrench, and whether some competitors might instead behave irrationally. The answers marked a shift from August, when management said irrational pricing was already happening in some segments.

Q: "Do you think that maybe creates some potential for irrationality just because some competitors may try to get market share that could put pressure on profitability of the system? I mean, we've seen some pressure on your fees, which I think you're doing it to protect the client base, but maybe there's further competitive pressures that could put some pressure on your overall profitability, right?"
— Tito Labarta, Goldman Sachs

A: "My thoughts on this question is, first of all, I believe that there is no room for irrationality. This is my first comment over here. Why is that? Because everybody who is operating in a relevant basis in the market know that the delinquency is going up of the system, the level of capital, it's being consumed, the excess capital, everybody doing their homework in terms of capital adequacy."
— Milton Maluhy Filho, CEO

Assessment: The competitive bear point has eased, and for an uncomfortable reason: competitors are expected to behave because their own capital and credit are under pressure. That is good for Itaú's pricing and a sign of how stretched the system is. On share, management repeated that share is a consequence, not a target, and that it gave up card share by design.

Margin and return on equity into 2027

A question on revenue asked whether the secured mix would keep compressing the margin in 2027 and what that means for return on equity. Management said the mix shift will continue, that it targets a stable risk-adjusted margin rather than a high headline one, and that it will not guide return on equity.

Q: "Question is, people are looking at 2027, and I wanted you to discuss the dynamic of revenue. We think about the mix more collateralized. It has led us to a drop in the margin. Looking up ahead, looking at 2027, the mix should not have any changes. Large corporate may be better, but it should be collateralized. Can you tell us on how do you understand the level of margin NII?"
— Marcelo Mizrahi, Bradesco BBI

A: "I believe that the NIM, the dynamic of the margin is not going to change. It's going to continue to grow in collateralized products. That generates this effect. [...] Talking about ROE, our expectation is to continue to deliver good profitability looking ahead. If the ROE will be higher or lower, we are working with this so we can calibrate what we expect, and it will depend on the scenario, specifically in the cost of credit that we will see looking at 2027."
— Milton Maluhy Filho, CEO

Assessment: The headline margin will keep drifting lower by design, and management would not say whether returns rise or fall next year. For a stock priced on the level of its return, an explicit refusal to call the direction is the most informative answer on the call.

Whether slower growth means a larger distribution

The last capital question asked whether lower 2027 loan growth would release more capital for dividends, or whether caution would argue for retention. Management restated the policy and leaned toward distribution.

Q: "We can expect a growth that is lower of the credit portfolio that inevitably will have more capital. Can you imagine this excess of capital being distributed in dividend or maybe cautious will require- This on the safe side. What you can expect in dividend with this scenario of more capital being used."
— Henrique Navarro, Santander

A: "It is difficult to say if it is going to be higher or less. We are going to have more BRL because last year we anticipated, jokes aside, specifically because we have a higher result this year. The dividend by share will be higher by consequence, and the level of payout will depend on all of these points. [...] I am not foreseeing anything that justifies a higher retention than what we have."
— Milton Maluhy Filho, CEO

Assessment: A higher dividend per share for 2026 is now close to a commitment, and the bank sees no reason to hoard capital in a harder year. That is the most constructive statement on the call for the equity, and the clearest support for the Bull 4 pillar.

What They're NOT Saying

1. The direction of 2027 returns: Asked directly, management said it could not say whether return on equity will be higher or lower next year, and gave no 2027 guidance. The answer arrives in February. Until then the Street is pricing a return that fell 0.7 points this quarter against a hurdle that rose 0.5 points.
2. The size of the forward-looking provision: Management confirmed that retail profitability in the quarter reflects a forward-looking scenario adjustment and said it is in "excess of comfort" on provisions. It did not size the adjustment, so investors cannot separate observed losses from scenario provisions in a cost of credit that rose 10.2% year on year.
3. How much of 2026 the tax revision is worth, and whether it lasts: The effective tax range moved 1.5 to 3.5 points lower. Management called it mechanical and tied it to the TJLP. It did not quantify the earnings effect or say whether the 28% to 29% range carries into 2027, when the rate path and the new consumption-tax regime both change.
4. What the higher hurdle means segment by segment: The nine-month business-model table shows credit earning 15.3%. The bank raised its cost of capital to about 15.25% in the same filing. It did not say whether credit allocation or pricing changes as a result, or restate value creation at the new rate.
5. The corporate exposure behind the Stage 3 coverage drop: Management attributed the fall in companies-side Stage 3 coverage to one fully provisioned client in bankruptcy leaving the book. It gave no size or sector, and companies-side Stage 3 coverage is down more than 8 points in a year.
6. The consumption-tax transition: The new CBS tax begins in two months. Management said much of its revenue falls under a specific regime and that it does not expect to pay more tax, but said it still has no clear view of the cash or accounting materiality.
7. What Avenue contributed, for a third quarter: The acquisition has been consolidated since the first quarter. It was not mentioned on the call, and its share of fee and revenue growth remains undisclosed, in a quarter when fee growth guidance sits at the floor.

Market Reaction

• Pre-print setup: The ADR closed at $7.05 on Tuesday 3 November, the day results were released after the close. It entered the print down 1.5% year to date, down 5.6% over twelve months and down 17.9% over the trailing 30 days, at the bottom of a 52-week closing range of $6.83 to $10.16 and 30.6% below the high. The S&P 500 was up 10.6% year to date on the same date.
• The October path: After the first round on 4 October, in which the opposition candidate led, the ADR rose 15.5% to $9.92 on 5 October and reached its $10.16 closing high on 9 October. After President Lula won the 25 October runoff with 51.79% of valid votes, it fell 20.4% on 26 October to $7.70, from $9.67. The local preferred share fell 13.7% that day and the real lost about 8% to 9% against the dollar. The other large Brazilian bank ADR fell 28.0% and the Brazil country fund 19.8%.
• Reaction session: On Wednesday 4 November, the day of the call, the ADR opened at $7.20, a 2.1% gap, traded between $7.07 and $7.30 and closed at $7.28, up 3.3%, two cents below the session high. The local preferred share rose 3.2% to R$39.71.
• Volume: 31.2m shares against a 30-day average of 40.5m, or 0.8 times. The average is inflated by election-week trading, which included a session of 137m shares on 26 October.
• Relative performance: The whole sector rose. Locally, the large private and investment-bank peers gained between about 3.7% and 6.0%, more than Itaú's 3.2%, and the Ibovespa rose 1.2%, ending a run of six straight losses that began with the first session after the runoff. The other large Brazilian bank ADR rose 4.0%, the Brazil country fund 1.6% and the S&P 500 1.6%.
• Since our initiation: From the $8.21 close on 5 August that anchored our August note, the ADR is down 11.3% against the S&P 500 down 0.5%.

The results did not move this stock in any meaningful sense, and the honest account of the last quarter is that the sovereign did. Between our initiation and this print the ADR rose 24% to its high on a first-round result the market liked, and then lost all of that and more on the runoff. Measured against that, a 3.3% gain on an in-line print is small. It came on a day when the US midterm results lifted risk appetite, Brazilian futures rates fell and every large Brazilian bank rose, most of them by more than Itaú. Unlike in August, the stock closed near its high rather than fading, which suggests the market found nothing in the call to sell. It also found nothing to make Itaú lead its sector.

Our Hold rating expects returns in line with the S&P 500, and over the last quarter the ADR has underperformed it by about 11 points. The reason is not in Itaú's numbers, which met every operating commitment, but the outcome still counts. Most of that underperformance arrived in a single session driven by a political event the rating did not anticipate. The more useful question is whether the repricing went too far, and that is a question about the cost of capital, which the bank itself just raised.

After the close on the reaction day, the central bank's rate committee held the policy rate at 13.75% in a unanimous vote, pausing a run of five consecutive cuts since March. The session traded before the decision, and we attribute no part of the move to it. The pause does matter for the outlook, because it is the "restrictive for longer" path management described on the call.

Street Perspective

Debate: Does Itaú still deserve a premium multiple if revenue growth stays moderate?

Bull view: The constructive camp argues that predictability is now the product. In a less predictable Brazil, a bank that forecasts its own delinquency to the decimal and earns 23.6% while its closest private peers reported 16.1% and 11.9% deserves to trade at a premium, and the premium is the price of quality rather than a bet on growth. Several desks kept the name as their preferred bank after the print.

Bear view: The sceptics note that recurring profit growth of about 6% year on year is the slowest of the year, that fee growth was close to nil, and that the sequential gain came from treasury income and a lower tax guide. On this reading the multiple already prices the superior fundamentals, and the question is how long the premium survives if revenue growth stays in the mid-single digits.

Our take: Both camps are right about different things, and the cost of capital decides between them. The quality is real and the relative gap to peers is widening. But a premium multiple on a mid-single-digit revenue grower is justified only by the spread of return over hurdle, and that spread narrowed by about 1.2 points this quarter. We would pay for the quality at about 1.8 to 2.0 times book, and the shares are at the top of that band.

Debate: Is the tax-rate cut an earnings upgrade or a low-quality offset?

Bull view: The constructive reading is that the tax revision is real money. It follows a higher long-term rate, it is mechanical rather than discretionary, and some estimate it adds about 1% to next year's earnings. Combined with strong treasury income, it means the bottom line has more support than the revenue lines suggest.

Bear view: The sceptical reading is that weaker client and fee revenue will absorb part of the benefit, that a tax rate is not a franchise, and that this is the second consecutive quarter in which tax guidance balanced softer revenue guidance. Few see the print driving material upward estimate revisions.

Our take: The bears have the better argument on quality and the bulls on 2026 arithmetic. We accept the 28% to 29% range for 2026 because management has guided it and the mechanism is clear. We do not carry it into 2027 until management guides it, because the same policy channel can move the other way.

Debate: After the election sell-off, is Itaú the defensive way to own Brazil, or the most exposed to a higher cost of capital?

Bull view: The prevailing view on the Street is that Itaú is the bank most prepared for a hard 2027: the strongest capital and liquidity, provisions recognised early, and a deferred-tax composition that carries no capital cliff. If weaker competitors retrench, it can take share at better prices, and a surplus-capital distribution in early 2027 pays investors to wait. On this view the October sell-off punished the best house with the rest of the market.

Bear view: The bear camp argues that the best balance sheet does not protect the multiple from the sovereign. The bank raised its own hurdle by 50bp in a quarter, its credit business now earns roughly its cost of capital, credit demand is already falling, and management expects a worse wholesale year. Defensive earnings can still be lower earnings.

Our take: Itaú is the right Brazilian bank to own, and that is a different question from whether to own it now. The relative case strengthened this quarter and the absolute case weakened, because the discount rate moved against all Brazilian equities. A confirmed distribution and a 2027 guide that implies a return near this year's level would tip us constructive. Until then the stock is a fair-value way to hold the Brazilian sovereign risk, not a discount to it.

Model Update

The table sets our August initiating assumptions against what the third quarter showed, and states where we now sit.

Valuation: The local preferred share closed the reaction session at R$39.71. On the June book value per share of R$18.92, the latest available, that is 2.10 times book. On the same basis, the R$42.38 close on 5 August, the reaction day that anchored our initiation, gave 2.24 times, so the local share has de-rated by about 6%. Tangible equity rose 6.7% in the third quarter, so on September book the multiple is likely closer to 2.0 times. Annualising the quarter's recurring result over the June share count gives earnings of roughly R$4.58 a share and a multiple of about 8.7 times on the reaction-day close, against about 9.4 times on the same basis in August. The 9.3 times we published in August used the pre-print close.

Valuation impact: A simple sustainable-growth framework sets a fair price-to-book multiple at the return less growth, divided by the cost of capital less growth. With the bank's own 15.25% hurdle, a 23.6% return and nominal growth of 5% to 7%, the framework gives roughly 1.8 to 2.0 times book. The same framework at August's 14.75% hurdle and 24.3% return gave roughly 2.0 to 2.2 times. The framework's band has fallen by about 9%, and the shares have de-rated by about 6%, so the sell-off has not opened a discount. On June book the shares sit just above the new band, and on September book at its top. We still do not publish a price target. The 2027 earnings base will be set by the guidance in February, and a target struck today would mainly be a view on the Brazilian sovereign.

Thesis Scorecard Post-Earnings

The pillars below are the same eight established at initiation, graded against what the third quarter showed. Status tags use ON TRACK, AT RISK or BROKEN for bull pillars, and CONTAINED, EMERGING or MATERIALISING for bear points.

August Commitments Revisited

Overall: The operating thesis strengthened and the investment case did not. Every operating commitment we set in August was met, one bear point eased, and the four bull pillars held. Against that, the bank raised the hurdle its returns are judged by, its return fell for a second quarter, its sequential growth leaned on trading, and one bear point moved from warning to forecast. The franchise is better than the price action implies, and the macro risk around it is higher than it was in August.

Action: Hold. We keep the triggers we published in August, and none of them tripped. The Underperform triggers were a second revenue guide cut, small-business delinquency above the guided 2.1%, a second decline in companies-side Stage 2 coverage, or a policy move against the sector's effective tax rate. The Outperform signposts were client margin at its floor on operating drivers, a stable fee line inside the new range, and confirmation of the early-2027 distribution. The first two Outperform signposts were met in form but helped by the calendar and guided to the edge, and the third is reaffirmed but not yet confirmed. We pre-state the triggers for the next two prints. Outperform: a confirmed early-2027 additional distribution (the extraordinary distribution of our August signpost) together with a 2027 guide that implies a return near this year's 24%, which would lift our fair band above the current price; or a fall in the shares below about 1.8 times book with the guide intact. Underperform: a 2027 guide that implies a return below 22%, a cost-of-credit guide above this year's R$43.5bn ceiling, or a further rise in the bank's cost of capital.