DLO — Thread 2026-08-10 (English)

1/12 dLocal's take rate collapsed from 2.79% to 0.84% in six years. The market reads that as slow death. But gross profit in dollars still grew ~40% YoY. Two facts, one company. Only one can be right about where this goes.

2/12 What DLO does: one API that lets global merchants (Amazon, Spotify, Uber) collect and pay across ~44 emerging markets. It handles 900+ local payment methods, 37 licenses, FX and compliance country by country. The messy last-mile nobody else wants.

3/12 Thesis #1: the market reads basis points, the business generates dollars. Take rate fell, but TPV compounds ~77%/yr. Gross profit still grew 37-40% YoY WHILE the rate dropped. A smaller cut of a vastly bigger base is still more money.

4/12 Thesis #2: a real regulatory moat. 37 licenses across 26 markets, 15+ years of last-mile compliance that Stripe/Adyen structurally avoid. It shows up in retention: net revenue retention 145%, volume retention 158%.

5/12 Thesis #3: fortress balance sheet. Zero debt, zero goodwill, ~$720M cash, self-funded via merchant float. Net invested capital is basically negative, so it earns profit on almost no capital. ROE 24-35%.

6/12 Anti-thesis #1: the moat protects volume, not price. A 0.84% take rate is forensic proof there's no pricing power. If it breaks 0.75% AND volume slows below 20%, this is a low-return utility, not a compounder.

7/12 Anti-thesis #2: it's a binary bet dressed as a compounder. Upside and downside are separated by ONE unknowable variable — the take-rate floor — which management admits it doesn't control ("output metric, not a target").

8/12 Variant: consensus prices a utility in decline at 8x gross profit. The read here is different — there's an economic floor to the take rate (~0.6-0.8%, Visa/MA-style) on a much larger base. The edge exists only IF the rate stabilizes.

9/12 Valuation goalposts: no-growth floor (EPV) ~$6. Base case ~$27. Bull ~$63. At ~$14, roughly 43% of the price is the floor — the market is paying ~57% for growth that may or may not show up.

10/12 Multiple: 8.8x EV/gross profit, bottom of its range. DLO grows GP faster than Adyen (~37% vs ~25%) but trades at ~8.8x vs Adyen's ~30x. Cheap for a reason, not by mistake — the reason is take-rate doubt.

11/12 What to watch in Q2 2026: (1) take rate — floor near 0.83-0.85% or breaking below 0.80%? (2) operating income/gross profit — back above 55% or stuck at 44.5%? (3) adjusted FCF recovering from a $15M collapse. One quarter isn't a trend.

12/12 Full pre-earnings breakdown — what to watch and what each outcome would mean — in the podcast: [PODCAST_LINK]. Not financial advice. Do your own research.