XP Inc.: A 22.5% ROAE and Excess Capital Meet a 1.20% Take Rate, and 2.11 Times Book Leaves No Margin of Safety

The second-quarter numbers show that barbell almost perfectly. Retail fixed-income revenue fell 16% to R$833m because of mark-to-market effects. Equities rose 11% to R$1.138bn, Funds Platform rose 23% to R$418m, Cards rose 16% to R$375m, Credit rose 27% to R$105m, Insurance rose 23% to R$80m and Other Retail rose 28% to R$813m, with XP identifying banking, float balances and newer businesses as contributors. Wholesale Corporate revenue more than doubled to R$606m, supported by credit, derivatives, FX and trading.

The result was respectable rather than explosive: second-quarter gross revenue rose 8% to R$5.056bn, net revenue 9% to R$4.884bn, gross profit 9% to R$3.353bn, pre-tax managerial income 15% to R$1.565bn and adjusted net income 5% to R$1.384bn. Gross margin was 68.6%; adjusted ROAE was 22.5%; adjusted ROTE was 27.2%. The company is generating enough profitability to absorb some take-rate pressure, but its revenue mix is becoming harder to analyze because a larger portion depends on market-making, spreads, credit and risk-weighted assets.

The quality of the balance sheet is better than the headline R$408bn asset total makes it look, but not as asset-light as XP’s historical market narrative implies. At June 30 the group had R$377.2bn of financial assets, including R$254.6bn at fair value through profit or loss and R$91.7bn at amortized cost. Loan operations were R$35.4bn. On the liability side, R$97.7bn related to retirement-plan and insurance liabilities that are substantially matched by policyholder assets and should not be treated like ordinary corporate debt. Funding instruments payable totaled R$116.7bn; deposits represented R$70.2bn and structured notes R$22.5bn within market funding.

Credit has not yet produced a visible deterioration commensurate with wholesale growth. Gross loan operations at June 2026 were about R$35.9bn, of which R$603m were Stage 3, an IFRS credit-impaired proxy of roughly 1.7%; Stage 3 balances had declined from roughly R$696m at December 2025. The loan ECL allowance was about R$423m, and second-quarter ECL expense was R$96m. These figures do not eliminate concentration or structured-credit risk, and Stage 3 is not identical to a Brazilian regulatory NPL ratio, but there is currently no evidence in the published credit table of the deterioration that would justify treating XP’s loan book as distressed.

Capital is unusually important to the present stock story. XP’s managerial BIS ratio was 20.3% in June versus its stated 16–19% target, with CET1 of 17.1%. Applying 20.3% to R$126.6bn of RWA gives about R$25.7bn of regulatory capital; at an 18% ratio on the same RWA, only R$22.8bn would be required. The mechanical excess is about R$2.9bn before allowing for future RWA growth, buffers and earnings. XP explicitly says it intends to move into its target range through capital distributions.

That makes repurchases a material part of per-share compounding. XP spent roughly R$996m buying treasury shares in the first half of 2026, after R$1.899bn of repurchases in 2025; the August cancellation retired another 11.79m shares. The H1 cash-flow statement also records R$517.9m of dividends paid. A new R$1bn buyback authorization remains in place. At today’s roughly R$103 translated share price, every R$1bn used to repurchase stock can retire about 9.7m shares before offsetting issuance. Against that, annual share-based compensation was R$590m in 2025, so gross repurchases substantially overstate the net shrinkage available to ordinary holders.

The share-count discipline also has to be viewed through governance. XP remains founder-controlled through XP Control LLC and ten-vote Class B shares. The Cayman structure and foreign-private-issuer status permit governance practices that need not mirror a US domestic issuer. XP has nevertheless added formal governance architecture over time; in 2024 it announced a Risks, Credit and ESG Committee and a Strategy and Performance Committee.

The governance discount became more tangible in 2025. On March 12 short seller Grizzly Research accused XP of improperly using derivatives and proprietary funds to create trading profits; XP called the allegations false, later sued Grizzly for defamation in US federal court and said the Gladius and Coliseu funds had no outside investors and the challenged transactions complied with Brazilian law. XP shares fell 5.5% on the day of Grizzly’s report. The existence of allegations is a fact; their truth is not established by the short report. The appropriate investment conclusion is that the episode raised the cost of opacity in a business already rich in derivatives, proprietary funds and related financial instruments, rather than that the allegations were proven.

Accounting reinforces that point. XP’s second-quarter managerial pre-tax income was R$1.565bn while accounting pre-tax income was R$1.515bn, yet both presentations ended with the same R$1.384bn net income. The R$50m difference came from classification between pre-tax lines and tax-related items, not from an adjustment that changed shareholders’ earnings. In the latest quarter, “adjusted net income” is economically the IFRS net-income number.

The unusually low tax rate deserves more skepticism than management’s earnings adjustments. XP’s accounting effective rate was only 5.1% in fiscal 2025, while its managerial normalized rate was 13.4%; the second-quarter managerial rate was around 12.4%. The financial-statement tax reconciliation shows large reductions arising from entities operating under different tax regimes and differential taxation of intercompany transactions. This makes a return to the full Brazilian headline corporate rate unlikely without structural tax reform, but I would not capitalize a 5% IFRS rate forever. A normalized low-teens rate is a more defensible valuation assumption.

At the September 18 close, the market capitalisation is R$52.34bn against June equity attributable to owners of R$24.83bn: 2.11 times book. Annualizing the R$1.384bn second-quarter net income produces R$5.54bn, so the stock is around 9.45 times annualized earnings. Those multiples look low beside US brokers. A Brazilian real cost of equity is far higher. Charles Schwab traded around 19.2 times trailing earnings on September 18, Interactive Brokers around 36.1 times and Robinhood around 53.0 times. The comparison illustrates why simply applying a global brokerage multiple to XP would be a category error.

The qualitative portrait is “company in transition.” XP has already proven that it can dismantle part of Brazil’s incumbent-bank investment-distribution economics. The question for the next cycle is whether it can turn that distribution advantage into a diversified financial institution while retaining platform-like returns on equity. That transition can create more stable revenues, but it also introduces balance-sheet, credit, capital and governance risks that the original brokerage did not carry.

The market is currently trading three things at once: the beginning of a lower-rate cycle, a potential recovery in risk-asset participation, and the release of excess capital through buybacks and dividends. A recent market commentary explicitly framed the rate cut as a catalyst for XP’s activity, although that commentary quoted R$2.2tn of client assets; I reject that number because XP’s own second-quarter disclosure says R$1.535tn. Primary disclosure wins.

The sharpest bull-bear disagreement is therefore not about whether XP can grow revenue next year. It is about what kind of ROE that growth earns. If a 20–24% ROE survives with a take rate around 1.2%, credit losses stay contained and the share count falls, today’s valuation can support good compounding. If wholesale expansion drives RWA materially faster than earnings while take-rate competition pushes retail economics toward 1.1%, the current 2.1 times book is expensive relative to a normalized BRL cost of equity in the high teens.

Vertical history and financial evolution

XP’s original opportunity arose from a peculiarity of Brazilian finance. Household savings were historically concentrated inside large universal banks, which controlled deposits, credit, product manufacture and distribution. An investor commonly bought whatever fund, CDB or investment product the relationship bank put on its own shelf. XP’s early model attacked distribution rather than trying to build another universal bank at once: financial education generated leads, brokerage provided market access, independent advisers supplied local relationships, and an open product shelf weakened the banks’ ability to keep clients captive. XP’s annual filings still describe the resulting business as an investment platform built around distribution and client assets.

Founded in 2001 under Guilherme Benchimol, the business initially had little resemblance to the present prudential conglomerate. The lasting insight was that advice could itself become a distribution channel. Rather than spend decades building branches, XP could let entrepreneurial adviser offices acquire clients locally while the central platform supplied products, custody, execution, technology and compliance. The model carried an inherent conflict: advisers historically had strong transaction and product-placement incentives. That conflict is one reason Brazil’s subsequent adviser-compensation and transparency reforms matter directly to XP’s take rate.

The first real stage was distribution formation. XP used education and brokerage to pull customers away from bank branches and gradually assembled what became an open-architecture marketplace. The durable capability created in this period was the ability to recruit advisers, supply them with products and convert household savings into platform assets, not software or low-price execution alone.

The second stage was institutionalization. Outside capital and acquisitions expanded the product shelf and gave XP the resources to challenge banks nationally. Itaú’s eventual investment was strategically revealing: an incumbent that had every reason to resist open architecture instead acquired a large economic interest in the challenger. The subsequent XPart restructuring and 2021 merger into XP distributed Class A shares to Itaú shareholders and unwound the earlier ownership relationship. By 2023 the old shareholders’ agreement among XP Control, General Atlantic and Itaú-related entities had been terminated; the platform had become institutionally independent of Itaú even though Itaú remained one of its most formidable competitors. XP’s filings document the Cayman holdco and historical shareholder arrangements.

The third stage was public-market acceleration. XP Inc. was incorporated in the Cayman Islands in 2019 and listed Class A common shares on Nasdaq that December. The issuer is XP Inc.; the core Brazilian operating businesses sit underneath it, including XP Investimentos and Banco XP. The public-market story initially emphasized a technology-enabled, asset-light challenger taking share from Brazil’s concentrated banks. That framing was reasonable for the distribution engine then, although it encouraged investors to compare XP with fintech platforms more readily than with balance-sheet financial institutions. SEC materials confirm the Class A structure and Nasdaq issuer.

The pandemic and very low Brazilian rates amplified that model. When risk-free returns fell, households had more incentive to seek funds, equities and alternative investments; an adviser-led open platform captured both net inflows and higher-monetization products. The market accordingly treated XP as a structural growth company. XP’s historical investor-relations chart shows the subsequent reversal in market perception: the post-listing rerating gave way to a steep derating as Brazil entered a much higher-rate environment, followed by partial recoveries rather than a return to the old valuation regime.

The fourth stage, from the 2022 rate shock through roughly 2024, forced XP to answer a harder question: what happens to an investment marketplace when the easiest product to sell is a government-linked fixed-income instrument with a double-digit yield? Client assets could still grow, but the mix moved toward lower-turnover and often lower-take-rate products. Equity brokerage weakened. XP responded by building businesses that use the client relationship beyond investing: cards, credit, insurance, retirement, banking services and a larger corporate and institutional franchise. The shift reduced reliance on retail trading but made the group more capital intensive.

By 2025–26 the fifth stage, hybridization, was visible in the financial statements. In fiscal 2025 gross revenue reached R$19.447bn, 8% above 2024, while adjusted net income rose 15% to R$5.218bn. Retail still generated R$14.584bn, but corporate and issuer services, institutional activities and new retail verticals were increasingly meaningful. Equities revenue fell 5% in 2025; fixed income rose 12%; retirement plans rose 20%, credit 24%, insurance 47% and Other Retail 22%.

†Approximate year-end level inferred from the company’s disclosed 2025 growth rate; the other figures are directly reported.

The ratios show the important financial trend better than the revenue growth does. Net income is growing faster than gross revenue because margin discipline, lower financing expense and the tax structure are helping. But client assets are growing faster than retail revenue, so monetization per unit of assets is compressing. That trade-off defines the present model.

The second-quarter take-rate decomposition supports three explanations. Mix is clearly important: fixed-income revenue fell despite large fixed-income client balances, while asset appreciation and inflows increased the denominator. Competition and pricing are also likely contributors because incumbents and new platforms can subsidize distribution to win affluent customers. Yet the sequential move from 1.18% to 1.20% suggests the decline is not simply a straight-line price collapse. The evidence is more consistent with a mix-driven structural reset toward roughly 1.15–1.25%, with pricing pressure preventing a return to the unusually rich 2020–21 economics.

XP’s adviser network gives the company more ways to respond than a conventional online broker. Transactional advisers monetize product sales and trading but carry the strongest conflict risk. Fee-based advisers can charge against assets and produce steadier economics with better alignment, although a transparent fee can be lower than embedded product economics. Registered-investment-adviser-style relationships push further toward recurring advice and away from transaction incentives. The transition matters because fee-based assets may lower the headline take rate while raising retention and earnings visibility. XP reports the aggregate adviser count but does not disclose enough publicly to model the profitability and assets of all three channels separately. The disclosure gap is an important one.

Retirement may be the most strategically attractive bridge between the old and new model. XP reported R$101bn of retirement-plan client assets in Q2, up 18% year on year, while proprietary XP Vida e Previdência assets reached about R$97bn, up 34%; XP estimated a roughly 5% share of Brazil’s PGBL/VGBL market. Retirement assets are sticky, long-duration and naturally suited to advice. They also generate the R$97.7bn retirement and insurance liability visible on the balance sheet, which is largely paired with policyholder assets rather than representing conventional leveraged borrowing.

The wholesale-bank pivot is more consequential. Corporate revenue of R$606m in Q2 rose 117%, with XP pointing to credit, derivatives, FX and trading. Institutional revenue was R$383m, up 12%, while Issuer Services fell 30% to R$186m. That contrast tells the cycle story: issuer services depends heavily on whether capital markets are open, while corporate derivatives and credit can earn revenue even when IPO or debt underwriting is quieter. The downside is capital consumption. RWA rose 26% in a year, faster than both revenue and earnings.

“Expanded loan portfolio” of R$77.9bn should not be confused with accounting loans on the balance sheet. IFRS loan operations were R$35.4bn at June 30. The broader managerial figure includes exposures beyond plain on-balance-sheet loans. The distinction is essential when evaluating leverage and credit loss.

Credit quality is currently benign enough to let management continue expanding, but the wholesale book has not yet crossed a full Brazilian credit downcycle at its present scale. Gross IFRS loan operations were about R$35.9bn, Stage 3 about R$603m and total loan ECL allowance about R$423m. Stage 3 represented roughly 1.7% of gross loans. XP wrote off about R$141m in the first half. These numbers do not reveal obligor concentration, collateral quality or the loss distribution inside structured products, so they should be read as early warning gauges rather than a complete risk map.

The securities balance is even less amenable to a simple interest-rate sensitivity. XP had R$206.9bn of securities at fair value through profit or loss and R$47.7bn of derivative assets, offset in part by R$42.3bn of derivative liabilities and other funding and hedges. Applying a gross “duration × yield change” calculation to the R$207bn securities number would grossly overstate directional exposure because the book contains trading inventory, hedged positions, client facilitation and financing structures. XP does not disclose a consolidated DV01 that lets an outside investor cleanly translate a 100-basis-point rate move into earnings. The observable evidence is instead the R$155m year-on-year decline in Q2 retail fixed-income revenue that management explicitly tied to mark-to-market.

Funding has also evolved away from the simple “broker platform” image. At June 2026 XP showed R$116.7bn of financing instruments payable, including R$70.2bn of deposits, R$19.2bn of financial bills and R$22.5bn of structured notes, plus repo funding. This is still different from a traditional bank funded principally through granular checking and savings deposits, but the platform has become a powerful captive distribution channel for its own funding instruments. That is a genuine strategic advantage as long as customers regard XP paper as an appropriate part of their portfolios.

The dollar debt is declining. The first-half cash-flow statement records R$2.286bn of debt-securities repayment, while the June 30 balance of debt securities had fallen to R$2.466bn from R$5.037bn at year-end. This is consistent with settlement of the 3.25% senior notes due around July 2026, leaving the 6.75% 2029 dollar bond as the principal international senior issue. I did not find a separate repayment notice in the primary materials retrieved, so I would not claim a more precise funding bridge; the accounts indicate internal liquidity and ordinary platform funding rather than issuance of another similarly sized dollar bond.

A conventional industrial-company free-cash-flow analysis is inappropriate here. For a financial institution, movements in repos, trading assets, deposits, loans and funding instruments all run through the cash-flow statement and can overwhelm the economic cash generated by the franchise. In H1 2026 operating cash flow was R$3.526bn versus roughly R$2.70bn of net income, a ratio near 1.3 times, but that does not mean earnings had “130% cash conversion” in the industrial sense. Purchases of PPE were R$88.8m and intangibles R$261.6m, only R$350m combined in six months, so physical and software investment is not the binding capital claim; regulatory capital against RWA is.

For the owner-earnings check required in valuation, I therefore use net income minus an estimated maintenance portion of physical/intangible investment, while separately charging growth for regulatory capital. Annualizing H1 PPE and intangible additions gives roughly R$701m. Assuming 60% is maintenance, an explicitly judgmental estimate because XP does not disclose the split, maintenance investment is about R$420m. Against R$5.536bn of annualized Q2 net income, that gives owner earnings around R$5.12bn, or approximately R$10.06 per current total share. The current owner-earnings P/E is around 10.2 times, modestly above the 9.45 times headline annualized P/E. The gap is far below 30%, so accounting earnings remain usable for valuation after the regulatory-capital adjustment.

The history of XP’s valuation has moved through three identities. Public markets first paid for a high-growth, capital-light disruption story. The high-Selic period forced the multiple down as transaction revenue weakened and investors recognized macro sensitivity. The current share price embeds a third identity: a high-ROE Brazilian financial institution that may return substantial capital but should command neither an uncritical fintech multiple nor a low-quality bank multiple.

Business model, moat, cycle and competitive landscape

XP makes money by sitting between Brazilian wealth and financial products. That sentence sounds simple; the current revenue engine is not.

Retail remains the economic core. In Q2 2026 it generated R$3.881bn of the R$5.056bn gross revenue. Equities contributed R$1.138bn, Fixed Income R$833m, Funds Platform R$418m, Retirement R$118m, Cards R$375m, Credit R$105m, Insurance R$80m and Other Retail R$813m. Wholesale generated the remaining R$1.175bn.

Source: XP Q2 2026 results; percentages are calculated from reported gross revenue.

The cost base has enough fixed infrastructure to generate operating leverage, but compensation remains meaningfully variable. Q2 SG&A excluding D&A was about R$1.64bn. People expense was R$1.109bn; salaries and payroll taxes rose 23%, bonuses rose 12%, while share-based compensation fell 32% to R$110m. LTM compensation ratio was 23.2% and the efficiency ratio 34.3%. Headcount rose 13% year on year to 8,491, faster than active clients.

Scale does not automatically improve margins. Technology and central infrastructure are scalable, but growth in wholesale banking, control functions and employed advice adds people and regulatory costs. The long-term operating-leverage case requires revenue per employee and revenue per adviser to rise, not merely more personnel.

The first real moat is distribution. Eighteen thousand-plus advisers create a human network that is expensive to reproduce and gives XP reach beyond what a pure app can accomplish with affluent households. Distribution also feeds the balance sheet: the same client base can buy funds, bonds, XP-issued funding instruments, pension products, cards, insurance and credit. That is a stronger moat than the app itself.

The second is product breadth and open architecture. XP became relevant because clients could access a wider shelf than the traditional captive-bank model. As incumbents copied open shelves and digital investing, openness itself stopped being unique. What remains defensible is the combination of shelf breadth, adviser workflow, custody, market access, research, product structuring and a recognized investing brand.

The third is scale under regulation. The prudential conglomerate, broker-dealer, bank, insurer, asset manager and insurance broker sit under Central Bank and CVM supervision. Licenses are not an absolute barrier to entry, but the cost of compliance, risk systems, capital, custody infrastructure and adviser oversight favors institutions already operating at scale. The same regulatory system can nevertheless compress XP’s economics by forcing more transparent adviser remuneration and reducing conflicts embedded in product commissions. CVM Resolution 178 is part of the modern regulatory framework for investment advisers, and the regulator has continued public guidance around the regime.

The fourth moat is switching friction rather than contractual lock-in. Portfolios contain pensions, funds, structured products, financing arrangements, cards, collateral relationships and adviser ties. Moving everything is possible, but inconvenient. The NPS trend shows that this friction cannot be mistaken for unconditional loyalty: XP’s annual report showed NPS falling from 72 in 2023 to 70 in 2024 and 65 in 2025; Q2 2026 recovered slightly to 66.

That is why I classify the adviser network and regulatory scale as real moats, while “technology platform” is mostly a marketing moat. Brazilian competitors have enough engineering talent and capital to reproduce a good app. It is much harder to reproduce eighteen thousand advisers, a trillion-real-plus asset base, a product-manufacturing network and the trust required to fund a bank through the same channel.

The industry itself remains attractive because XP still has room to take wallet share rather than needing Brazil’s savings pool to grow extraordinarily fast. The 2025 annual report put XP at about 12% of an R$8.6tn investment market. Total addressable assets are not the constraint. The constraint is whether each incremental real of assets carries enough fee, spread or ancillary-product income to earn above the cost of the capital and people required to serve it.

XP sits at the intersection of three cycles. The first is the Selic cycle, which sets the attractiveness of risk-free products and banking spreads. The second is the capital-markets cycle, which determines trading, equity activity and issuer fees. The third is the credit cycle, increasingly relevant as Corporate and lending grow. In the original XP, the first two dominated; in the current XP, all three matter.

Brazil entered the report date early in an easing cycle. Selic at 13.75% remains extremely high in nominal terms. The Focus consensus at September 14 expected 12% in 2027 and 10.5% in 2028, while the Central Bank remained cautious because inflation was not yet fully anchored. The survey expected year-end 2026 IPCA around 4.9% and USD/BRL around 5.20.

The October 2026 general election increases the variance around that path. Fiscal credibility directly affects the real, inflation expectations and the term structure of Brazilian interest rates; those variables then affect XP twice, first through its securities and funding books and second through client willingness to leave risk-free assets. On September 18 the real closed near R$5.144 per US dollar, and Brazilian markets reacted to renewed fiscal concerns despite the rate cut.

This produces a counterintuitive scenario. A responsible post-election fiscal path combined with falling inflation would probably be XP’s best macro environment: lower Selic, lower long-end yields, stronger BRL, higher equity valuations, more capital-market issuance and gradual migration toward higher-take-rate assets. A disorderly fiscal outcome could keep Selic elevated or reprice long yields higher; XP would keep some float income but suffer mark-to-market volatility, weak issuer activity and lower investor risk appetite.

The direct competition is best understood as several different businesses converging on the same affluent Brazilian wallet.

BTG Pactual is the closest high-quality financial comparator. It became an integrated investment bank, asset and wealth manager with a sizeable credit and markets franchise. Its appeal to wealthy and institutional customers is the depth of its investment-banking and product-manufacturing machine rather than a mass adviser network. In Q2 2026 BTG reported roughly R$5.1bn of profit, a 26.7% ROAE and about R$2.7tn of assets under management or administration; the market nevertheless reacted cautiously to faster credit expansion.

XP versus BTG is a contest between distribution-first and manufacturing/banking-first models. XP historically owned the adviser channel and sourced products from many manufacturers; BTG historically owned more of the institutional and balance-sheet engine. Both have moved toward the other. If XP can generate BTG-like returns while preserving a more open consumer distribution proposition, its multiple deserves support. If wholesale growth merely makes XP a less efficient investment bank, BTG is the better model.

Itaú became the opposite type of competitor: a universal relationship bank with enormous deposit, credit and client-data advantages that has learned to offer more open investment products. The main customer reason to choose Itaú is convenience and institutional trust: salary account, card, mortgage, corporate relationship, investments and private banking can sit together. XP’s reason to exist was that this integrated model historically did not offer the best investment shelf. As Itaú’s investment proposition improves, XP must win through advice and choice rather than simply being “more open.”

Nu Holdings attacks from below. Nubank’s mass-market distribution, low servicing cost and primary-account engagement are much broader than XP’s affluent-investment heritage. At the end of 2025 Nu had about 131m customers across Brazil, Mexico and Colombia and was still growing rapidly. Its immediate threat to XP is owning the financial relationship before a younger customer becomes affluent, not complex wealth advice.

Inter is a smaller version of the same convergence. Its digital banking and investment ecosystem can cross-sell to a broad customer base, but its US-listed market value was only about US$2.35bn, or R$12.1bn at the report-date exchange rate, versus XP’s R$52.3bn. Its smaller balance sheet means less direct scale today, but it remains a useful indicator of how low-cost digital banking can bundle investments without relying on a huge external adviser network.

The US brokers are useful for business-model anatomy, not for direct valuation. Schwab shows how enormous client assets plus cash balances can create rate sensitivity alongside brokerage and advisory fees. Interactive Brokers illustrates the economics of automated, global, low-cost execution. Robinhood shows what the US market can pay for rapid retail engagement and product expansion. On September 18 their trailing P/E ratios were roughly 19.2, 36.1 and 53.0 respectively. XP’s 9–10 times current earnings is not automatically “cheap” against them because XP earns BRL cash flows under a much higher nominal cost of equity, carries Brazilian political and currency risk, and has controlling ten-vote shares.

Nu and Itaú illustrate the same problem with market-cap comparisons. Their September 18 US-listed market capitalisations were about US$65.3bn and US$80.4bn, respectively, dwarfing XP’s roughly US$10.2bn, but each capitalizes a very different earnings stream and risk profile. Nu is priced for international digital-bank growth; Itaú for a large universal-bank franchise; XP for a hybrid wealth platform and investment bank.

XP’s ecological niche remains valuable: it is the independent distribution leader sitting between the universal banks and the mass digital banks. BTG is most likely to take high-end investment and corporate economics; Itaú can win by bundling the whole relationship; Nu can intercept future affluent customers before they need an adviser. XP wins when customers actively care about investments as a category and want advice plus product choice. That is a narrower niche than “all financial services,” but a defensible one.

Current fundamentals, governance and regulation

The latest quarter shows a company whose headline growth rate understates the internal rotation.

Gross revenue grew only 8%, but Corporate grew 117%, Funds Platform 23%, Other Retail 28%, Credit 27% and Insurance 23%. Fixed Income fell 16% and Issuer Services fell 30%. Net inflow of R$28bn was much stronger than the R$10bn year-earlier quarter, with retail inflow R$20bn, up 28%. Over the trailing twelve months, XP attributed R$103bn of asset growth to net inflows and R$60bn to market appreciation.

That decomposition matters for forecasting client assets. A simple base case starts with R$1.535tn. If net inflows settle around 6–7% of opening assets annually and markets add 3–4%, client assets can grow around 9–11% without extraordinary share gains. A bull case with a successful rate-cut cycle, healthier equities and stronger fixed-income mark-to-market could reach 12–14%. A bear case with election stress and weak markets can reduce appreciation to zero or negative even if XP continues gathering assets.

The take rate then determines whether that asset growth becomes retail revenue. At 1.20%, R$1.535tn theoretically represents an annualized revenue pool around R$18.4bn if the metric were applied mechanically to all assets, but XP’s reported take rate is a managerial measure with mix and averaging conventions, so it should not be used as a direct revenue multiplication formula. The better forecasting discipline is to model the direction: 10% asset growth with a take rate falling 5% produces only about 4–5% monetization growth before newer verticals.

I see limited evidence that XP can return sustainably to a 1.33% take rate without a major equity-market boom. Fee transparency, incumbent competition, a larger fixed-income share and migration toward fee-based advice all work against that. The base case holds the take rate near 1.18–1.22% over the next cycle. A move below 1.15% would be a significant negative signal because the burden of growth would move even more heavily onto wholesale and banking.

The wholesale story deserves a higher evidentiary bar. R$606m of quarterly Corporate revenue is a large number, but RWA growth of 26% is faster than group revenue growth of 8%. XP does not disclose a clean segment ROE showing whether incremental Corporate capital is earning 20% or 30%. Until it does, investors should not equate revenue growth with value creation.

At the group level, capital gives management room. BIS was 20.3%, CET1 17.1% and RWA R$126.6bn. Management’s 16–19% target effectively says it believes the current capital buffer is excessive. At the 18% midpoint, the static June excess is about R$2.9bn. At 19%, it is about R$1.65bn; at 17%, about R$4.18bn.

My base distribution model assumes RWA grows around 10% a year rather than the current 26%, because a continuation of 26% would eventually collide with both capital and risk appetite. I assume H2 2026 earnings of R$2.8bn, 2027 earnings of R$6.0bn and 2028 earnings of R$6.6bn. Holding BIS around 18%, the purely mechanical distribution capacity is about R$3.4bn for H2 2026, R$3.8bn in 2027 and R$4.1bn in 2028. These are not forecasts of announced payouts: they are upper-end capital arithmetic before management buffers, acquisition needs, stress capital and model uncertainty.

A more prudent payout range is R$2.5–3.0bn in H2 2026, R$2.8–3.4bn in 2027 and R$3.2–3.7bn in 2028. That still implies unusually high distributions relative to the R$52.3bn market cap. I would favor buybacks over dividends below my base intrinsic-value range and dividends once the stock approaches or exceeds it. At R$103, a R$2bn buyback retires about 19.4m shares gross. R$590m of annual SBC at the same price represents roughly 5.7m share-equivalents, leaving net shrinkage around 13–14m shares, or 2.5–2.8%, if issuance tracks accounting compensation. Actual issuance prices and vesting terms will differ.

The accounting balance sheet contains several items that investors should adjust mentally rather than simply subtract.

Deferred tax assets were R$3.829bn gross at June 30, with roughly R$1.379bn related to tax-loss carryforwards; net deferred tax assets after liabilities were around R$3.20bn. Gross DTA equals about 15% of equity attributable to owners, so realizability matters. XP’s sustained profitability makes a large write-off unlikely in my base case, but a severe earnings downturn would affect both current profits and the value of part of those assets.

Goodwill and intangibles were R$2.954bn, about 12% of equity. That level is noticeable but not dominant. Investments in associates and joint ventures were R$3.718bn; about R$2.214bn were equity-accounted and R$1.504bn represented fair-value associates. These stakes add optionality and accounting opacity at the same time.

The R$97.7bn retirement and insurance liability should largely be viewed as pass-through policyholder economics rather than ordinary financial leverage. Treating every liability on the R$408bn balance sheet as creditor leverage would materially misread the business.

The managerial-versus-IFRS reconciliation is less alarming than it first appears. For Q2, IFRS service revenue included brokerage commissions, securities placement, management fees, insurance brokerage and other commissions; financial-instrument income made up the rest. Accounting PBT was R$1.515bn versus managerial R$1.565bn, but each arrived at R$1.384bn of net income. I therefore use IFRS net income for valuation and treat the managerial presentation as a segmentation tool.

The discontinued managerial “Other” revenue line and the 2026 transfer of Institutional into Wholesale reduce historical segment comparability. XP has also changed how fund tax withholding is classified. Investors comparing old slide decks to new ones can create false growth or margin signals if they do not restate the business perimeter. The accounting statement is the anchor when the managerial labels move.

Tax is the more material comparability issue. FY2025 accounting income before tax was about R$5.45bn and net income R$5.17bn, implying an accounting ETR around 5.1%; management’s normalized adjusted ETR was materially higher. In H1 2026 the statutory reconciliation showed substantial deductions tied to entities taxed under different regimes and to intercompany taxation. The Cayman parent’s tax status is part of the legal structure, but the evidence does not support saying “Cayman explains the low tax rate.” Brazilian subsidiary and fund regimes do much more of the work.

For valuation I normalize tax to roughly 12–15%. A legislative change eliminating important regime differences or deductions could therefore reduce my earnings estimates, but I am already refusing to capitalize the 5% accounting rate indefinitely.

Governance requires a similar distinction between structural risk and allegation.

Structurally, Class B carries ten votes per share, XP Control holds the controlling Class B block, and the Cayman foreign-private-issuer structure provides fewer US-style governance constraints than a one-share-one-vote domestic issuer. XP Control was created as part of the controller reorganization and its activities are restricted principally to holding XP shares.

The February 2026 controller realignment concentrated ControlCo around Guilherme Benchimol and a smaller group of senior XP figures while several long-time partners exited economically. The September conversion of a block of Class B shares for resale changes the Class A/Class B mix but not the company’s total economic share count. The key minority-holder fact remains that roughly one-fifth of the economics can carry around seven-tenths of the vote. The governance discount should be treated as permanent until the dual-class structure meaningfully unwinds.

Management turnover compounds that concern modestly. CEO Thiago Maffra had to cover the finance role temporarily after CFO Victor Mansur left in 2026 before Gustavo Viviani, a long-tenured Santander Brasil executive, took over in August. One CFO transition does not establish weak controls, but two finance-leadership handoffs in a short period deserve monitoring in a group whose balance sheet and derivatives activities are becoming more complex.

The 2025 short-seller episode is the obvious meaning behind management’s language about strengthened governance and controls. Grizzly alleged improper proprietary-fund and derivative economics; XP denied the claims and went to court. The legal action does not itself prove XP’s accounting, just as the short report did not prove fraud. For shareholders, the appropriate response is to demand better visibility into proprietary funds, related transactions, derivatives and segment capital returns.

CVM rules are both moat and threat. Licensing, adviser registration, suitability, product governance and disclosure raise the operating cost for a new entrant. Rules increasing compensation transparency make it harder for platforms to hide economics inside product commissions. Over time that favors firms that can earn an explicit advice fee or monetize the broader customer relationship, while pressuring the old transaction-heavy take rate. CVM continues to treat the investment-adviser regime as a formal regulatory priority.

Brazilian prudential rules play the same dual role. They permit Banco XP and the group to turn distribution into credit, derivatives and funding economics, but growth then consumes regulatory capital. The company’s own 16–19% BIS target is more useful to this equity analysis than a generic Basel statutory floor because it incorporates management’s desired operating buffer. A fall below 16% would be a warning; a persistent ratio above 20% after management’s year-end commitment would signal inefficient capital allocation.

For foreign investors, the final risk is translation. A Nasdaq holder owns a claim on BRL earnings but sees a USD share price. The share can appreciate in BRL while producing a mediocre dollar return if the real depreciates. At September 18, US$20.01 translated to R$102.93 at BRL5.144/US$. A 10% weaker real with unchanged BRL intrinsic value reduces dollar value by roughly 9%.

Valuation, risks and catalysts

The right starting point is book value and normalized earnings.

June equity attributable to owners was R$24.831bn. Dividing by 508.50m total economic shares gives book value around R$48.83 per share. The R$102.93 translated market price equals about 2.11 times book. Annualized Q2 net income gives about R$10.89 of earnings per current share and a 9.45 times P/E. Using the maintenance-investment owner-earnings adjustment described earlier gives roughly R$10.06 per share and a 10.2 times owner-earnings P/E.

The multiple is not cheap in isolation because XP’s nominal BRL equity hurdle is high. I derive the base cost of equity from a normalized Brazilian short-rate path rather than the spot 13.75% Selic alone. The September Focus path averages roughly 10.8% across 2027–29; I add a 5.5–6.0 percentage-point equity and country-risk premium plus a small governance/FX premium, producing a base nominal BRL cost of equity around 17–17.5%. Conservative and optimistic cases use roughly 19% and 16%. The risk-premium inputs are my valuation assumptions; the Selic path is market consensus.

A useful sanity check comes from the bank Gordon model:

Justified P/B = (ROE - long-run growth) / (cost of equity - long-run growth)

At a 22% sustainable ROE, 17% cost of equity and 6% long-run book-value growth, justified P/B is only about 1.45 times. To justify roughly 2.1 times book at a 17% cost of equity while holding 22% ROE, the formula needs long-run growth around 12–13%, a demanding assumption for a company simultaneously distributing excess capital. Alternatively, investors must believe the true cost of equity is closer to 15%, the sustainable ROE exceeds 22%, or book value materially understates economic franchise value.

This explains the apparent contradiction between a 9.5 times P/E and a 2.1 times P/B. The P/E looks low because XP earns a high ROE on a relatively small equity base. The P/B says the market already expects a substantial portion of that high ROE to persist. The stock is therefore inexpensive on earnings but not obviously cheap on capital.

Historical percentile claims would be false precision. XP’s public history is short and the business mix has changed sharply since IPO; pre-2022 “platform” P/E multiples and present “hybrid financial institution” P/B multiples are not like-for-like. The economically relevant statement is that the market has permanently lowered the multiple from the early growth-platform regime while still assigning XP a sizeable premium to book.

Peer valuation reinforces rather than solves this problem. Schwab’s roughly 19 times, IBKR’s 36 times and Robinhood’s 53 times trailing earnings all embed lower home-currency discount rates or much stronger growth expectations. XP’s discount is justified in part by BRL rates, macro volatility, wholesale credit risk and control. A convergence to US broker multiples is not part of my base case.

BTG is a more relevant economic benchmark. Its 26.7% Q2 ROAE is above XP’s 22.5%; that helps explain why investors can rationally award BTG a premium despite its larger balance-sheet exposure. XP must show that its distribution franchise can lift group returns rather than watching the wholesale pivot dilute them.

The absolute valuation rests on the following scenarios. These are valuation-scenario analysis within a research framework, not investment advice.

†Approximate three-year return from the 2026-09-18 price, including modeled cash dividends; buybacks are reflected through per-share terminal values. These are assumptions rather than forecasts from company guidance.

The conservative case is not a disaster scenario. It assumes XP remains profitable, client assets continue rising and ROE remains near 20%, but take-rate pressure and capital consumption prevent a high valuation. The base case requires no return to 2021 conditions: roughly 10% asset growth, a take rate around 1.2%, normalized low-teens taxes, controlled RWA expansion and meaningful repurchases are enough.

The optimistic case needs several things to work simultaneously. Selic declines without fiscal disorder; Brazilian risk assets deepen; retail take rate recovers above 1.24%; wholesale grows without credit losses; and ROE stays around 24%. The valuation then deserves a stronger premium to book.

The expectation gap is concentrated in four figures: take rate, RWA growth, ROE and share count. The next earnings print matters less for gross revenue than for whether those four move together in a favorable direction. A quarter with 10% revenue growth but 25% RWA growth and another take-rate decline would be poor. A quarter with 7% revenue growth, stable take rate, RWA growth near 10–15% and a 2% lower share count would be better.

The margin-of-safety check is less flattering than the headline P/E.

Current R$102.93 is above the conservative R$92–100 intrinsic-value range. The discount to conservative value is zero. This is not a stock priced for a bad outcome.

The most fragile base-case assumption is sustainable ROE above 21% while take rate remains around 1.2%. If the incremental value attributed to above-cost-of-equity returns is cut to 70% of my base assumption, the base fair value falls toward roughly R$105–112. That range is very close to the current price.

If earnings were flat for three years, capital distributions would still provide some return, but the result would be materially below the base-case 17.5% BRL CAGR and likely in the mid-single-to-high-single digits depending on RWA. That does not beat the present 13.75% Selic hurdle. A buyer today therefore needs growth, successful capital return or a lower future discount rate; flat earnings alone do not create enough compensation.

Margin-of-safety sufficiency verdict: none.

The risk that matters most is structural take-rate erosion. I assign medium probability and high impact. The observable warning is a retail take rate below 1.15% for two consecutive quarters while client assets continue growing. The transmission path is direct: revenue grows slower than assets, adviser and technology costs cannot fall proportionately, ROE declines and the justified P/B compresses.

The second is wholesale capital misallocation, medium-to-high probability and high impact. The current warning already exists: RWA grew 26% year on year. The risk becomes material if RWA keeps growing above 20% while group ROE falls below 20% or Stage 3 loans rise above roughly 3%. Revenue would still look healthy initially, but required capital would absorb earnings that otherwise could be distributed.

The third is Brazilian fiscal and election risk, medium probability and high impact for a dollar holder. A fiscal shock raises the long end of the curve, weakens BRL and delays rate cuts. That can hit fixed-income mark-to-market, client risk appetite, issuer services and the USD translation simultaneously. The September 18 move in the real and equities after renewed fiscal concern is a small illustration of the channel.

The fourth is governance and conduct risk, low-to-medium probability but high impact. The observable indicators are CVM or Central Bank enforcement, adverse findings around proprietary funds or related transactions, another abrupt senior-finance departure, or evidence that the controller uses its voting power in a way that disadvantages Class A investors. The 2025 Grizzly litigation raises the disclosure bar even though it does not establish wrongdoing.

The fifth is tax normalization, medium probability and medium impact. A move from my modeled 13% tax rate to 20% cuts after-tax earnings by roughly 8% on the same pre-tax profit. The market could absorb that if ROE and flows are strong; combined with take-rate pressure it would materially weaken the earnings case.

The sixth is hidden dilution. Buybacks are attractive only to the extent shares actually disappear. Investors should compare dollars or reais spent with the total Class A plus Class B count, not simply accept a repurchase headline. The August cancellation is positive evidence because it permanently removed 11.79m treasury shares.

The positive catalysts over the next 12 months are concrete: continued Selic reductions without BRL disorder, a retail take rate holding at or above 1.20%, RWA growth decelerating toward the low teens, completion of enough distributions to bring BIS inside 16–19%, and risk-asset activity reviving issuer and equity revenues. The first rate-cut catalyst has already begun.

Negative catalysts are equally observable: an election-driven fiscal shock, fixed-income mark-to-market losses, Stage 3 migration, another quarter of 20%+ RWA expansion, take rate below 1.15%, or a regulatory event around advice and product conflicts.

†Q3 had not ended on the research base date, so no Q3 2026 result existed. XP’s IR site lists Q2 2026 as the latest reported quarter; mid-November is the expected reporting window rather than a company-confirmed day retrieved in this research.

The dashboard should be read as a system. A falling take rate is manageable if net inflows are strong, efficiency improves and buybacks shrink the denominator. Fast RWA growth is manageable if ROE rises and Stage 3 remains controlled. A lower BIS ratio is desirable when caused by distributions, dangerous when caused by losses. No single metric carries the thesis.

Cross-synthesis and final research conclusion

XP has proven one capability beyond reasonable doubt: it can change the way Brazilians buy investments.

That achievement was partly an era effect. Falling transaction costs, digital distribution, declining trust in captive bank shelves and periods of low Selic all helped. But it was not merely luck. Building a national adviser network, scaling client assets beyond R$1.5tn and forcing incumbent banks to respond required execution. The historical moat was distribution innovation.

The question today is whether that capability transfers into banking.

There is a strong economic reason to try. A wealth platform has limited monetization if every asset becomes a cheap fixed-income product. The client relationship is worth more if the same household uses a card, takes credit, buys insurance, keeps retirement money on the platform and finances XP’s balance sheet. Likewise, corporate clients can be served with derivatives, credit, issuance and institutional distribution. The strategy raises revenue per relationship and makes the franchise less dependent on equity trading.

The cost is that the business begins to look like the institutions XP originally disrupted. Capital must be held. Credit can go bad. Trading books generate mark-to-market noise. Funding matters. Regulators care about prudential ratios, not only user growth. Senior management needs bank-grade risk systems. The 26% rise in RWA is thus one of the most important numbers in the entire report.

The transition still looks economically rational because current group profitability is good. A 22.5% ROAE at a point when retail take rate is only 1.20% is evidence that XP is no longer dependent on the unusually favorable transaction environment of its early public years. Corporate, Other Retail, funds, insurance and credit are compensating for weaker parts of the old engine.

Yet the market is already paying for persistence. At 2.11 times book, the valuation needs ROE to remain substantially above a normalized BRL cost of equity for years. That is different from saying “9.5 times earnings is cheap.” At a 17% nominal cost of equity, 22% ROE creates value; at 18% ROE, the excess return becomes thin. This stock can therefore move sharply even if earnings do not decline much: a change in confidence about sustainable ROE can reprice book value from two times to something closer to one-and-a-half times.

Horizontally, XP’s strongest advantage over BTG, Itaú, Nu and Inter is that investments are the center of its customer relationship. BTG starts from capital markets and wealth manufacturing; Itaú from universal banking; Nu and Inter from the primary digital account. That focus gives XP credibility with customers who actively care about investing and gives its adviser channel something competitors cannot cheaply recreate.

Its weakness is also visible in that comparison. Itaú and Nu can treat investments as part of a much broader relationship and accept lower direct investment monetization. BTG can monetize sophisticated clients through investment banking, market-making and asset management. XP is therefore squeezed from both directions if investment distribution itself becomes commoditized.

The strategic answer is fee-based advice plus broader wallet share. That can work, but investors should accept that the take rate may never return to historical highs. A healthy XP in 2030 may have a lower retail take rate but higher revenue per client, more recurring advice, more retirement assets and a larger wholesale contribution. Judging management by whether take rate returns to 1.33% would miss the point. Judging it by whether ROE stays above 20% while take rate stays near 1.2% is better.

The retirement business supports that transition more convincingly than many newer verticals. R$101bn of retirement assets are sticky and fit the advice proposition. Cards also reinforce the daily relationship, with Q2 TPV of R$13.5bn and around 1.6m active cards. Credit and corporate banking are more ambiguous because they produce revenue by consuming scarce capital.

Capital return is the near-term bridge from business quality to shareholder return. Management has told investors that 20%+ BIS is temporarily excessive. The static excess over an 18% ratio is about R$2.9bn, and ongoing earnings replenish capital. Even after allowing for RWA growth, several billion reais per year of dividends and repurchases are plausible under the base case. That is enough to make per-share earnings grow meaningfully faster than group net income if buybacks remain disciplined.

The best capital-allocation test is price-sensitive behavior. At R$75, buying XP shares would be a highly attractive use of excess capital under my assumptions. At R$200, the company should distribute cash instead of repurchasing aggressively. Management’s willingness to change the mix will tell minority holders whether the buyback program is genuine capital allocation or merely a mechanism to neutralize compensation dilution.

The 2025 short-seller episode does not overturn the investment case, but it changes what evidence should be demanded. A business with R$250bn-plus of fair-value financial assets, large derivatives books, proprietary funds and an expanding wholesale bank deserves more granular disclosure than a simple brokerage. XP’s decision to sue Grizzly is a forceful response, but the best long-term answer is transparent segment capital, credit concentration and proprietary-trading disclosure.

The same is true of taxes. I do not believe the right bear case is “XP will suddenly pay 34%.” The financial statements show why the consolidated rate differs from the headline statutory rate. The more prudent assumption is that a 5% accounting ETR contains benefits too favorable to capitalize forever, while a low-teens normalized rate is defensible.

For the next year, the critical variables are take rate, post-election rates, RWA growth and distributions. The share price could rerate quickly if Selic keeps falling, the real remains stable, take rate stays near 1.2% and management takes BIS below 19% through buybacks. Conversely, the same period can expose the flaws of the hybrid model if long rates rise, wholesale credit expands too quickly and client assets remain trapped in low-monetization fixed income.

Over three years, the decisive variable is sustainable ROE. If XP settles at 21–23%, the current business model works. If ROE falls to 16–18%, a large portion of today’s premium to book disappears even if nominal earnings remain positive.

Over five years, the decisive variable is whether XP becomes the default independent financial relationship for affluent Brazilians rather than merely their investment account. Retirement, advice, banking, cards, credit and insurance need to deepen retention without turning the group into a mediocre universal bank. That is the strategic test that separates successful diversification from empire building.

The dollar investor has an additional hurdle. My base case assumes USD/BRL around 5.20 by 2029, close to the latest Focus expectation for the nearer term. If the real instead moves toward 5.8, a respectable BRL return can become nearly flat in dollars. That is why the conservative scenario produces about 5% annualized in BRL but only about 1% in USD. Currency is part of the investment result, not an accounting footnote.

The market may be misjudging two things in opposite directions. It may underestimate how much excess capital XP can distribute if RWA growth normalizes. But it may also underestimate how much of the current 22% ROE is required merely to justify a two-times-book valuation under Brazil’s cost of equity. The first supports the stock; the second limits the margin of safety.

Core bull reasons

• XP is still gathering assets: Q2 total net inflow reached R$28bn and client assets reached R$1.535tn, while the annual report suggests XP still controls only about 12% of Brazil’s investment market.
• A 22.5% ROAE with a 1.20% retail take rate shows that newer revenue sources are already offsetting part of the compression in the original brokerage model.
• A 20.3% BIS ratio versus the 16–19% target creates several billion reais of potential capital return when combined with continuing earnings.
• The August cancellation of 11.79m treasury shares shows that repurchase spending can become permanent per-share accretion rather than remaining treasury stock.
• A controlled Selic-cut cycle can improve equities, fund flows and capital-market activity even as some banking float income falls.

Core bear reasons

• Retail take rate has fallen materially from the end of 2024, so client assets can continue rising without producing comparable revenue growth.
• RWA is growing 26%, much faster than group revenue, and XP does not disclose enough segment capital data to prove that the wholesale expansion earns its cost of capital.
• The stock already trades around 2.11 times book even though a normalized BRL cost of equity is high, leaving no discount to my conservative value.
• Founder control through ten-vote shares, foreign-private-issuer exemptions and the 2025 proprietary-fund controversy justify a persistent governance discount.
• A dollar investor carries BRL risk on top of the operating thesis; fiscal stress can simultaneously hurt the currency, rate path, mark-to-market book and risk-asset flows.

Pre-mortem. One plausible 50% loss script starts in 2027. Itaú and BTG become more aggressive in fee-based affluent advice while Nu graduates more clients into investments. XP’s take rate falls through 1.10% and eventually toward 1.05%, yet wholesale management keeps RWA growing near 20%. A Brazilian slowdown pushes Stage 3 loans above 3.5%, ROE falls to 16–17%, and investors reprice the stock from roughly 2.1 times book to 1.2–1.3 times. Even with positive earnings, that combination can halve the equity value.

A second script is macro-governance driven. Fiscal policy after the 2026 election unanchors long-term inflation expectations; USD/BRL moves above 6, long yields reprice upward and fixed-income mark-to-market remains adverse. At the same time, a regulator requires tougher changes around adviser conflicts or proprietary product disclosure. Net inflow halves, NPS falls below 60 and the market begins valuing XP as a controlled Brazilian bank rather than an investment platform. A 1.2–1.4 times book multiple on impaired book-value growth again produces loss on the order of 40–50%.

The evidence does not currently make either pre-mortem the central case. Stage 3 is low, flows are healthy, ROE exceeds 20% and capital is abundant. They matter because each script describes a path to permanent capital loss rather than temporary price volatility.

My final judgment is that XP is a good financial franchise at a price that already requires the franchise to remain good. The current P/E understates that requirement because book value and cost of equity expose it more clearly. Investors buying R$102.93 are effectively betting that a 20%+ ROE, roughly 1.2% take rate, disciplined wholesale capital and substantial distributions can coexist.

I think that combination is achievable, but the conservative valuation does not provide a margin of safety. The stock belongs in the hold zone rather than the ideal-entry zone. A fall toward R$74–80 without corresponding deterioration in ROE, credit or governance would change the equation materially; alternatively, several quarters of 20%+ ROE, sub-15% RWA growth and a falling share count could raise my conservative value enough to justify purchasing at a higher market price.

【Company-profile scores】

• Fundamental quality: high
• Growth: medium
• Moat: medium
• Financial soundness: strong
• Management credibility: medium
• Valuation attractiveness: medium
• Risk level: medium
• Suitable investor type: long-term growth / value / cyclical

【Investment rating】

• Rating: Hold
• One-line thesis: High-ROE distribution and capital returns offset take-rate pressure, but 2.1 times book leaves no conservative margin of safety.

【Ideal Buy Price】74–80 BRL Basis: at least a 20% discount to the R$92–100 value implied by the conservative scenario.

• Acceptable hold price: R$98–150, derived from approximately ±15% around the R$115–130 base-case value range.
• Clearly overvalued price: R$187–205, beginning roughly 10% above the upper end of the R$150–170 optimistic value range.
• Current-price classification: acceptable hold.
• Whether to wait for a better price: yes for a new position seeking a genuine margin of safety; R$74–80 is the preferred entry range unless operating evidence raises conservative intrinsic value first. The opportunity cost is missing a rate-cut/capital-return rerating while Selic falls.
• Target holding horizon: 3–5 years.
• Expected annualized return: conservative about 5.0% BRL / 1.0% USD; base about 17.5% BRL / 17.1% USD; optimistic about 28.8% BRL / 31.7% USD.
• Max-loss risk: roughly 40–50% in a combined take-rate, credit and multiple-compression event resembling the pre-mortems above.
• Reassessment triggers: retail take rate below 1.15% for two consecutive quarters; adjusted ROAE below 18%; Stage 3 loans above 3%; RWA growth remaining above 20% without higher group ROE; BIS remaining above 20% after the promised optimization period, or falling below 16% because of losses rather than distributions.

【Valuation Range】

• current: 102.93 BRL (close as of 2026-09-18)
• bear (conservative · ideal buy zone): [74, 80]
• base (fair · acceptable hold zone): [98, 150]
• bull (optimistic · above the clearly-overvalued line): [187, 205]

Research uncertainties. First, XP does not disclose enough wholesale segment capital and ROE to prove the incremental economics of the fastest-growing business. Second, the company does not provide a consolidated duration/DV01 measure that would make the R$207bn FVPL securities book directly rate-sensitive in a model. Third, the exact mix of assets and economics in transactional, fee-based and RIA advice is insufficiently disclosed. Fourth, I found accounting evidence consistent with repayment of the 2026 senior notes but not a separate primary repayment notice specifying the funding source. Fifth, I did not retrieve every post-February 2026 Schedule 13D/13G amendment, so the 20-F ownership table plus the disclosed September Class B conversion should be treated as the best available control picture rather than a forensic beneficial-ownership reconstruction.

Source hierarchy. The core financial analysis uses XP’s Q2 2026 earnings release and interim IFRS financial statements, XP’s FY2025 Form 20-F and investor-relations materials, and SEC-filed governance documents. Macro inputs use the Central Bank’s Selic framework, the September Focus consensus as reported contemporaneously, and current Brazilian market reporting. Market quotations use XP IR and dedicated market-data feeds. The short-seller controversy is described through Reuters because it reports both the allegations and XP’s legal response.

Other tickers mentioned

• BPAC11.SA: BTG Pactual is the closest Brazilian hybrid wealth, investment-banking and balance-sheet comparator.
• ITUB.US: Itaú Unibanco is XP’s large universal-bank competitor for affluent investment relationships.
• NU.US: Nu Holdings is the mass-digital challenger that can capture customers before they migrate into wealth products.
• INTR.US: Inter is a smaller digital-bank and investment-platform convergence competitor.
• SCHW.US: Charles Schwab is the global reference for scaled brokerage, advice and client-cash economics.
• IBKR.US: Interactive Brokers is the reference for automated low-cost global brokerage economics.
• HOOD.US: Robinhood illustrates the valuation attached to high-growth retail financial engagement in the US.
• BSBR.US: Santander Brasil is relevant to Brazilian banking competition and is the prior employer of XP CFO Gustavo Viviani.

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