DLO — dLocal Limited · Newsletter 2026-08-10

dLocal's take rate has fallen every year for six years, from 2.79% to 0.84%. Read in isolation, that looks like a business being commoditized toward zero. Read next to the volume, it looks like something else entirely. Both readings draw on the same filings, and only one is right about where the company is headed — with Q2 2026 shaping up as the first real referee.

dLocal is a cross-border payments company built for emerging markets. Global merchants — the likes of Amazon, Spotify and Uber — plug into a single API to collect from and pay out to users across roughly 44 countries, and dLocal handles the part nobody else wants: 900-plus local payment methods, 37 regulatory licenses, non-convertible currencies, settlement and compliance, country by country. It earns a small cut — the take rate — of total payment volume, or TPV, the dollars flowing through its rails. The tell of the model: on every $100 of TPV it keeps about $0.84 of gross profit. A small slice of an enormous plate.

The case rests on three things. First, the market reads basis points while the business generates dollars: the take rate keeps falling, but TPV compounds around 77% a year, so gross profit in dollars still grew 37-40% year over year even as the rate dropped — a smaller cut of a far bigger base is still more money. Second, a genuine regulatory moat: 37 licenses across 26 markets and 15-plus years of last-mile compliance that Stripe, Adyen and PayPal structurally avoid, which shows up in net revenue retention of 145% and volume retention of 158%. Third, a fortress balance sheet: zero debt, zero goodwill, around $720M of cash, funded by merchant float rather than borrowing, with net invested capital close to negative — the company earns its profit on almost no capital, at a return on equity of 24-35%.

On valuation the goalposts are wide, because they hinge on one variable. A no-growth floor — earnings power value — sits near $6 a share. The base case lands around $27, the bull near $63. At roughly $14, about 43% of the price is that floor, meaning the market is paying the other 57% for growth that may or may not arrive. On multiples, DLO trades at 8.8x EV to gross profit, the bottom of its own history, versus Adyen near 30x — despite growing gross profit faster. Cheap for a reason, not by mistake, and the reason has a name: the take-rate floor.

The risks deserve the same conviction as the thesis. The dominant one: the moat protects volume, not price. A 0.84% take rate is forensic proof there is no pricing power, and if it breaks 0.75% while volume slows below 20%, dLocal reveals itself as a low-return utility and value converges toward that $6 floor. The second: concentration. The top ten clients are about 61% of revenue and two exceed 10% each — a single renegotiation can move the whole model, as one did in early 2024. The early-warning signal to watch is a take rate under 0.80% for two or three straight quarters paired with gross-profit growth below 20%.

The honest label for DLO is a special situation with an asymmetric, uncertain payoff — a name to track quarter by quarter, not a buy-and-forget compounder. What to watch next is narrow and specific: whether the take rate steadies near 0.83-0.85%, whether operating income recovers above 55% of gross profit from a stalled 44.5%, and whether adjusted free cash flow rebounds from a $15M collapse. The full pre-earnings breakdown — what to watch and what each outcome would mean — is in the podcast: [PODCAST_LINK].

Not financial advice. Do your own research.