dLocal Limited DLO

The price asks normalized earnings to compound +9.5% a year for a decade. Volume compounds at +62%, gross profit at +33% — and normalized earnings have not compounded at all since 2023. The dive is the distance between those three lines, and its name is the take rate.

1. Net take rate — THE CONTESTED DRIVER. Gross profit / TPV: 2.17% (FY2021) → 1.91% → 1.56% → 1.16% → 0.99% (FY2025) → 0.84% (TTM), 0.72% in Q2 2026. At +60% TPV growth, each 10 bp of take rate is ≈$55M of annual gross profit on the TTM volume base.
2. TPV compounding. +61.5%/yr FY2021→FY2025; +92% in Q2 2026; FY2026 guide raised to +60–70%. Local-to-local flows are 61% of TPV — the highest-retention, lowest-take mix.
3. H2 2026 operating leverage. The held operating-profit guide (+27.5–32.5%) requires $163–175M of H2 operating profit vs $118M in H2 2025 — automation delivery and the end of front-loaded spend, graded at the next two prints.
4. Capital return actually executing. $86.1M of buybacks at a $12.50 average in H1 2026 plus a formalized 30%-of-FCF dividend; $213.9M of authorization remains against a Mar 19, 2027 expiry.

🐻 1. The take rate is the business, and it halves every ~3 years. 2.17% → 0.84% in five years with no floor demonstrated; Q2 2026 printed 0.72%. Management's "flat ex-one-merchant" framing concedes the mechanism: the merchants who drive growth price the platform, and there is always a next largest merchant.
🐂 Bull answer: Gross profit dollars compounded +32.6%/yr through the entire compression, and the mix now shifting (pay-ins, frontier markets, merchant-of-record and buy-now-pay-later products) carries higher takes. Concession: if Q3 prints below 0.68%, the floor story is management's word against five years of arithmetic — checkpoint 1 is registered as the falsifier.

🐻 2. Normalized earnings have not grown since FY2023. $192.0M → $66.8M → $166.6M → $159.9M (TTM). The bar (+9.48%/yr) asks for a decade of compounding a line that has never compounded across any two-year window on the normalized definition — SBC tripled and "non-recurring" items recur on schedule.
🐂 Bull answer: The FY2024 trough was an Argentina/Egypt float-and-rates shock inside the normalized window, and FY2026 guidance is the inflection, set by a team whose guides have been beaten or raised for two years. Concession: the guide is a promise about H2, and H2 must deliver +38–48% operating-profit growth — the checkpoint set grades it, not this page's prose.

🐻 3. The reported-cash mirage cuts both ways. The same float mechanics that made FY2024 FCF negative make today's TTM reported owner earnings ($362.6M) look like a −1.92%/yr bar. And the normalization is the company's own definition, unaudited as a standalone measure.
🐂 Bull concession: Conceded on the framing; that is why the governing base is the normalized line, the reported bar is printed as a memo with its warning attached, and checkpoint 5 tests adjusted cash delivery quarterly. Cumulative convergence is the long-run audit.

🐻 4. Concentration is the moat's price and its risk in one line. Two merchants each >10% of revenue; Brazil + Mexico + Argentina = 50.4% of revenue; ride-hailing drove the Q2 sequential jump. A single Tier 0 renegotiation or wallet shift (Egypt's shape, 2025) moves the P&L by more than a year of buybacks.
🐂 Bull answer: 760+ enterprise merchants, net revenue retention 153%, share of wallet still only low-teens. Concession: retention is a volume metric; it has never yet protected the unit price.

🐻 5. Insiders sell, the controller structure is permanent, and the big holder is leaving. All Class B votes sit with the founder group; Kanovich sells under a 10b5-1 plan; General Atlantic filed to sell ~$136.9M in Aug 2026. The 2024 buyback record — $101.1M repurchased in FY2024 before/at the start of the 2024 $200M program window, $0 in 2025, and that program expired in May 2025 with no repurchases made under it during 2025 — is what capital allocation looked like before the current CEO's second year.
🐂 Bull answer: The 2026 program is executing into weakness at $12.50 average prices with shares cancelled, the dividend is policy-formalized, and the Sep 2025 board overhaul is the control discount being addressed in the open. Concession: none of it changes the Class B math; the alignment sub-grade and the grade cap carry it.

🐻 6. Frontier credit is a habit, not an accident. Lending at 7%/15% to Aza Finance ended with a recoverability write-down of an FVPL asset (current portion $24.5M → $0.4M in six months) and an acquisition-by-call-option of the borrower's assets.
🐂 Bull answer: The absolute dollars are small (~1% of EV), the exposure was disclosed in the interim notes, and the acquired assets extend the African footprint. Concession: the disclosure arrived as a reassessment, not as a strategy — the candor sub-grade carries it.

DLocal connects global enterprise merchants to emerging-market consumers through one API and one contract: local payment acceptance (cards, Brazil's PIX, India's UPI, Kenya's M-PESA and 1,000+ other methods), payouts, and foreign-exchange (FX) conversion and settlement across 60+ countries, holding 38 licenses and authorizations. Revenue is processing fees plus FX spreads, earned on TPV. Mix is shifting toward local-to-local flows (61% of TPV in Q2 2026) — volume where DLocal never converts currency and earns the thinnest take — and toward a handful of Tier 0 merchants (ride-hailing, e-commerce, streaming, remittances) whose scale buys them volume-tier pricing. Roughly 760 enterprise merchants use the platform; net revenue retention was 153% in Q2 2026, the fifth straight quarter above 140%.

The quarter in one line: management raised the volume and gross-profit guides while holding the profit guide — the whole thesis in one print.

The rows that decide the thesis

Growth & profit ($M)

Cash ($M)

Count (M shares)

*Jun 30, 2025 comparative from the interim statements. n/d = Adjusted FCF was not disclosed on this basis before FY2023 in the filed record pulled. FY2021 capex includes $46.1M of intangible-asset purchases (filed). FCF = operating cash flow − capex; owner earnings = FCF − SBC; normalized owner earnings = Adjusted FCF − SBC. Delivered compounding: revenue +45.5%/yr and gross profit +32.6%/yr FY2021→FY2025; TPV +61.5%/yr; normalized owner earnings −6.9%/yr FY2023→FY2025.

DLocal guides full years, in growth rates, and the Arnt-era pattern is consistent: FY2024 TPV guided +40–50%, delivered +44% (in range). FY2025 was guided in Feb 2025 at TPV +35–45% / revenue +25–35% / gross profit +20–25% / Adjusted EBITDA +20–30% — then raised in Aug 2025 — and delivered above every initial line (TPV +60%, revenue +46.6%, gross profit +36.6%, Adjusted EBITDA +47%). FY2026 was set in Mar 2026 (TPV +50–60%, gross profit +22.5–27.5%, operating profit +27.5–32.5%), held at Q1, and raised at Q2 (TPV +60–70%, gross profit +25–30%; operating profit held with the drag named on the call). One structural honesty marker: management pre-commits to take-rate compression inside its guides rather than letting it arrive as a surprise.

Health verdict: unlevered in any economic sense — net cash of $727.6M on the bridge definition — but the balance sheet is a settlement machine, not a vault: receivables and payables each roughly doubled in six months with TPV, and the cash sits across 60+ jurisdictions' operating subsidiaries. Filed escrow/guarantee balances in Other assets include $24.6M of bank/regulatory escrow and restricted cash subject to regulatory requirements in Egypt and Morocco — small against the cash line, and the reason "trapped cash" is a working-capital risk here rather than a balance-sheet fact. Level-3 haircut sensitivity: not triggered — no private/Level-3 mark is deducted in the EV bridge.

Method: solve on enterprise value (EV); 10% required return; 10-year window; 2% terminal growth; terminal value = year-10 owner earnings × 12.75. Archetype ruling: the float/financials exception is OUT. DLocal's float is settlement float that turns in days and its economics already land in the P&L; the distortion is a normalization problem. The governing base is normalized owner earnings = Adjusted FCF (the company's filed merchant-working-capital-stripped measure) − SBC. Cross-check: the FY2022–FY2025 average of reported owner earnings is $168.0M, within 5% of the normalized TTM base — the normalization removes timing, not economics.

EV bridge ($M): market cap 288.4M shares × $14.67 = $4,231.1M + financial liabilities $64.6M + lease liabilities $2.7M − cash $794.9M = EV $3,503.5M.

Shrink, labeled. The named demonstrated input, computed in the model from the two filed diluted weighted-average share counts — DEMONSTRATED_SHRINK = 1 − 301,742,797 / 305,136,290 = +1.11% — is not a repurchase shrink: FY2025 repurchases were $0 ($101.1M was repurchased in FY2024 (executed before/at the start of the 2024 program window, per the equity statement); the 2024 $200M program expired in May 2025 with no repurchases made under it during 2025.) and basic weighted-average shares rose +0.33% (290,014,019 → 290,969,221); the diluted decline is fewer dilutive instruments, not fewer shares. No demonstrated credited bar is recognized; the credited rows above are ASSUMED scenarios. The H1 2026 delivered pace (6,887,815 shares cancelled in six months) would sit near the 2% row if repeated for a decade — one half-year under a program expiring Mar 19, 2027 earns no governing status. No guided count path exists, so there is no GUIDED bar.

Competitive advantage period (CAP): at +15%/yr normalized owner-earnings growth the price is justified in ≈5.4 years; at +20%/yr in ≈4.0 years; at +25%/yr in ≈3.0 years. Base rate: payments processors sustaining double-digit owner-earnings growth for a full decade are rare; DLocal's own delivered normalized growth is negative across FY2023→TTM. The bar sits between the two delivered truths — far below volume/gross-profit compounding, far above normalized-earnings compounding. Sensitivities (normalized / reported): hurdle 9% → +7.47% / −3.62%; hurdle 11% → +11.37% / −0.33%; terminal growth 1% → +10.44% / −1.17%; terminal growth 3% → +8.38% / −2.80%.

What you must believe (breakeven grid, $/share): revenue compounding (rows) × year-10 owner-earnings margin on revenue (columns); the margin ramps linearly from today's normalized 11.8%. Cells at or above $14.67 are bolded. The price is earned at ~5%/yr revenue growth with a 20% year-10 margin — a margin the normalized base has not printed since FY2023 — or at flat revenue with a 30% margin, near the float-flattered reported margin of 26.7%.

• License and local-entity footprint Answered — 38 licenses/authorizations across 26 markets (16 more in process), local teams and acquiring in 60+ countries; a decade of regulatory build that cannot be bought quickly.
• Integration depth Answered — one API/one contract across 1,000+ payment methods; merchants expand country-by-country inside the same integration (net revenue retention 153%).
• Scale economics Answered — volume buys downstream pricing, FX liquidity and data; the flywheel claim is consistent with the gross-profit dollar record.
• Pricing power Flag — the take-rate series is the verdict: five straight years of compression, driven by the largest merchants' volume tiers.
• Counterparty independence Contested — two customers >10% of revenue each; contracts terminable on short notice; Adyen and Stripe push into the same corridors with larger balance sheets.
• Float/settlement trust Thin — the Aza Finance episode is the one blemish on the settlement-integrity record the moat ultimately rests on.

C+Grade C+, capped by guidance and alignment. Sub-grades: Guidance credibility C+ · Execution B+ · Candor B− · Capital allocation B · Alignment C+. The strong legs are capital allocation (framework now executing — $86.1M of H1 2026 buybacks at a $12.50 average, shares cancelled, $57.2M dividend paid) and execution; the cap is Guidance (the FY2024 profit guides were missed and the record behind the grade is release-verified, with verbatim transcripts for only 2 of 12 quarters) and Alignment (insider/10b5-1 and General Atlantic selling, zero buys).

People: Pedro Arnt, CEO (CFO from 2017; co-CEO Sep 2023; sole CEO Mar 2024). Guillermo Lopez Perez, CFO (on the Q1/Q2 2026 calls). Founders Andres Bzurovski (Chairman from Sep 2025), Sebastián Kanovich and Jacobo Singer hold the Class B shares (5 votes each) through their vehicles.

Guidance credibility (C+): two straight years of burdened initial guides beaten on every line (FY2025) or raised at mid-year (FY2025, FY2026); take-rate compression pre-committed inside the guides. The held FY2026 operating-profit guide, with its drag named on the call, is either candor or a pre-built excuse — checkpoint 4 grades it.

Execution (B+): TPV +60% in FY2025 and +92% in Q2 2026 with headcount roughly flat in 2026; licensing and geographic expansion delivered on sequence; the Sep 2025 governance overhaul (board departures, new compensation and governance committees, majority-independent transition, treasury-share cancellation) was executed as announced.

Candor (B−): take-rate mechanics are quantified on the calls (merchant-level bridge, ex-merchant flatness claim) and working-capital failures are named in the quarter they happen; against that, prior-year tax adjustments and loss items recur as "non-recurring," and the Aza recoverability reassessment arrived after the exposure was built.

Capital allocation (B): the 2026 buyback is real execution ($86.1M at a $12.50 average, shares cancelled) and the dividend is policy (30% of prior-year FCF); the 2023 program bought the drawdown. Debits: $101.1M was repurchased in FY2024 (executed before/at the start of the 2024 program window, per the equity statement); the 2024 $200M program expired in May 2025 with no repurchases made under it during 2025, and frontier counterparty credit (Aza) sits outside the core competence the capital should fund.

Alignment (C+): the cap. Class B control is total and permanent; insider flow is one-directional selling (Kanovich's small 10b5-1 parcels — 4,700 shares at $15.50 and 72,753 at $15.5369 in Aug 2026); the largest outside holder is trimming $136.9M into the company's own bid. No open-market insider purchase was located in the 2025–2026 Form 4 record. The company's buyback, not insider buying, is the alignment evidence.

Re-verification, Oct 17, 2026: grade reverted to C+ by the Oct 17, 2026 overnight-batch audit (see Corrections) — the record behind the grade is release-verified, with verbatim transcripts for only 2 of 12 quarters, the FY2025 20-F (Items 6/7), the Schedule 13D/A of Aug 19, 2026, the Form 144 of Aug 17, 2026, and the Form 4 record pulled from EDGAR. Verdict: C+ stands as corrected. Investor day: none in the last 24 months — verified negative; no CMD targets exist to score.

• Leverage — PASS. Net cash $727.6M on the bridge definition; financial liabilities of $64.6M are working-capital facilities, not term debt.
• Moat — FLAG. Pricing power fails the Fisher test (see Moat) and two merchants each exceed 10% of revenue on short-notice contracts.
• Valuation — PASS. On the normalized base the price embeds +9.48%/yr — demanding against normalized history, undemanding against every volume and profit-dollar line; no perfection is priced on either base.
• Management — FLAG. One-directional insider/holder selling and total Class B control keep the gate lit even with the guidance record as strong as it is.
• Capital allocation — FLAG. The lapsed 2024 program and the Aza credit episode are the two exhibits; the 2026 execution is the repair in progress.
• Accounting quality — PASS. IFRS statements with a filed non-IFRS reconciliation; the normalization used here is the company's own, printed with its bridge. Recurring "prior-year tax adjustments" are a named watch, inside Candor.
• Fraud / red flags — PASS with a named watch. The Muddy Waters-era securities class actions (first filed Feb 23, 2023) remain disclosed in the FY2025 20-F; no new accounting allegation is in the filed record since, and no restatement has occurred.
• Macro / sector — FLAG. Argentina + Egypt FX/rate economics, Brazil tax-regime evolution, Mexico tariff sensitivity (management's own guide caveats), and Pillar Two tax pressure from 2027.

Grading convention: an outcome between a checkpoint's PASS and FAIL lines is MIXED — neither passed nor failed — and the item is re-graded at the following print.

Update protocol: this page is re-scored against the lines above after the November 12, 2026 print; every threshold was set on Oct 17, 2026 and will not be redrawn after the fact. Re-basing triggers: a take-rate FAIL re-opens the governing base debate; a checkpoint 5 FAIL two quarters running retires the normalization to memo status and the reported bar governs.

1. Oct 5, 2026 edition — implied FCF bar of −4.2%/yr: retired. It was computed on reported FCF at a $13.74 price and priced float timing. The re-derived bars (normalized +9.48%/yr governing; reported −1.92%/yr memo) supersede it.
2. Oct 5, 2026 edition — "no debt": corrected. At Jun 30, 2026 the company carries $64.6M of financial liabilities plus $2.7M of leases against $794.9M of cash (net cash $727.6M).
3. Watchlist-era management grade C+: superseded by the grade in the verdict box, derived from the 12-quarter calls record, the guidance ledger, and the FPI insider/buyback record in the dossier.
4. Oct 17, 2026 — management grade reverted B− → C+ (sub-grades: Guidance credibility C+ · Execution B+ · Candor B− · Capital allocation B · Alignment C+), per the Oct 17, 2026 overnight-batch audit: the cap is Guidance (the FY2024 profit guides were missed and the record behind the grade is release-verified, with verbatim transcripts for only 2 of 12 quarters) and Alignment (insider/10b5-1 and General Atlantic selling, zero buys); the strong legs are capital allocation (framework now executing — $86.1M of H1 2026 buybacks at a $12.50 average, shares cancelled, $57.2M dividend paid) and execution. Class B votes corrected: Class B shares carry 5 votes each (FY2025 20-F), not 20 votes each as printed. 2024 buyback program corrected: $101.1M was repurchased in FY2024 (executed before/at the start of the 2024 program window, per the equity statement); the 2024 $200M program expired in May 2025 with no repurchases made under it during 2025 — not “expired unused” as printed.

• Transcript basis: full verbatim transcripts were read for 2026Q1 and 2026Q2; 2024Q4–2025Q4 rest on the filed earnings releases (which carry the guidance language delivered on those calls) plus call reporting; 2023Q3–2024Q3 on filed releases and contemporaneous reporting — the IR webcast archive retains one year. Page quotes are confined to the full-transcript quarters.
• Cash by jurisdiction: the filings pulled do not present a full country split of the $794.9M cash line; escrow and restricted amounts are filed, but a jurisdiction map of the whole balance is not — and is not estimated.
• Merchant-level economics: the two >10%-of-revenue customers are not named in the 20-F, and no per-merchant take rate is disclosed; the ex-one-merchant take-rate claim is graded as management's claim, not as a verified figure.
• FY2024 initial guide set: the TPV line is on file; the full initial set is not in the pulled record (dossier guidance ledger, seed gaps).
• Adjusted FCF before FY2023 was not disclosed on the current definition in the record pulled; normalized owner earnings therefore start at FY2023.

• T1 — FY2025 Form 20-F, filed Mar 18, 2026 (acc. 0002070979-26-000113): financial statements, Adjusted FCF reconciliation, share capital, major shareholders, buyback programs, subsequent events.
• T1 — Q2 2026 interim financial statements, Form 6-K filed Aug 13, 2026 (acc. 0001846832-26-000031): H1/Q2 2026 statements, balance sheet at Jun 30, 2026, Note 14 share capital.
• T1 — Q1 2026 Form 6-K, filed May 14, 2026 (acc. 0001846832-26-000019), incl. the Q1 2026 earnings release exhibit.
• T1 — Schedule 13D/A (General Atlantic), filed Aug 19, 2026: 36,814,136 Class A shares (~21.4% of Class A).
• T1 — Form 144 (General Atlantic DO B.V.), filed Aug 17, 2026: proposed sale of 9,842,559 Class A shares (~$136.9M).
• T1 — Form 4 (Kanovich), filed Aug 11, 2026: Aug 4–5, 2026 conversions and sales under a 10b5-1 plan adopted Nov 26, 2025.
• T2 — Earnings releases filed as 6-K exhibits: Q4 2024 (Feb 27, 2025), Q1 2025 (May 14, 2025), Q2 2025 (Aug 13, 2025), Q3 2025 (Nov 12, 2025), Q4 2025 (Mar 18, 2026) — guidance statements, quarterly Adjusted FCF, TPV.
• T2 — Earnings-call transcripts: Q1 2026 (May 14, 2026) and Q2 2026 (Aug 13, 2026), full text (MarketBeat); Q2 2026 prepared remarks (company IR).
• T2 — Q3 2026 print-date announcement, GlobeNewswire, Oct 14, 2026 (company release): Nov 12, 2026 after close, call 6:00 p.m. ET.
• T3 — Prices: Finnhub daily candles, corroborated by StockAnalysis (price data only). Analyst estimates: none used as inputs anywhere.