StoneCo Ltd. STNE

Key data

The business

Stone sells payment tools to Brazil's small and mid-size merchants: the terminal on the counter, a business bank account, working-capital credit, and software. It makes money three ways. It takes a cut of each card transaction, it charges monthly subscription and terminal rental, and it earns financial income, the interest on merchant deposits and on its own credit book. That third line now dominates: R$10.0B in 2025, 71% of revenue, because Brazil's benchmark rate has sat high and Stone keeps the spread on the float. The moat is distribution, a feet-on-the-street sales force working local hubs, plus the friction of moving a store's whole money stack. Stone's own 20-F concedes that friction is fading: PIX instant payments and open-finance rules have lowered switching costs, "making retention increasingly challenging." What you are really buying is a bank wearing a payments logo.

The numbers

The quarters tell one story: the payments engine is shrinking and the rate engine is carrying the total.

Transaction revenue has fallen every quarter, R$658mn to R$427mn, down 35% year on year, while financial income rose enough to hold the total flat near R$3.6B. Revenue per active client landed at R$251 in Q2, up 1.5% sequentially but down 6.6% on the year, so monetization per merchant is still eroding. Q2 adjusted EPS of $0.47 edged past the $0.463 consensus, the third small beat in four quarters. The R$300M rate headwind management flagged on the August call has not bitten yet: financial income actually rose 3.2% sequentially in Q2, the opposite of what the warning implied.

Over three years revenue grew a third, but the mix flipped: financial income went 6.2 to 10.0B while the card-swipe line fell 3.1 to 2.5B. This growth is rate-made, not payments-made.

At 4.2x earnings, 1.4x book, and a return on equity near 30%, Stone throws off cash it is using to retire stock at four times profit. Even flat operating income compounds hard per share when you shrink the count that fast. The catch, and the one thing the market and I are really arguing about, is quality: the engine doing the compounding is high-Selic float, which is cyclical, while the payments core it was built on keeps shrinking. The print that settles it is whether transaction revenue and revenue-per-merchant flatten before the rate cycle turns.

Management

Insider activity over the past year is small and roughly balanced: founder-linked Diego Ventura Salgado bought about $217K in May near current prices, and director Silvio Morais sold $102K in June, plan status not disclosed, so read it as a wash rather than a signal. The real capital story is the buyback, retiring stock at a mid-single-digit multiple, which is the most accretive thing a business this cheap and this cash-generative can do. That management is spending on its own shares rather than reaching for growth is the honest read on where they think the value is.

How it fails or surprises you

Take-rate erosion. The card-swipe line fell 35% year on year and ARPAC is down 6.6%. If PIX and open finance keep hollowing out card monetization faster than banking and credit fill the gap, the payments franchise melts while the rate income masks it. Watch transaction revenue and ARPAC over the next two quarters.

The rate turns. Financial income was 71% of 2025 revenue and rose again in Q2. When Brazil's benchmark rate falls, that spread compresses and the flat-total illusion ends. Watch financial income sequentially against the Selic path; a down quarter there is the tell.

Credit and buybacks compound (right tail). Stone is growing an unsecured merchant credit book while retiring stock near 4x earnings. If losses stay contained as the book scales and the count keeps shrinking, EPS compounds double digits on flat operating profit, and the market pays nothing for it today. Watch portfolio growth alongside delinquency and coverage.

Closing thoughts

This is a largely priced situation, not a hidden gem. At four times earnings the market is quoting a melting EM payments book with currency and political risk stapled on, and it is not wrong about the direction of the swipe line. My read differs on magnitude, not fact: the earnings are real, the balance sheet is fine at 1.4x book, and buybacks at this multiple are potent enough that flat profit still builds value per share. But the multiple-versus-peers gap is smaller than it looks once you mark the quality down for how much of it is Selic float rather than durable payments economics. The upside tail is modestly the fatter one at this price if credit holds, and the fatal tail is a loss cycle in the unsecured book arriving just as rates fall and demand softens.

The bet is still that Brazilian shopkeepers keep their Stone machines on the counter and keep their cash in Stone accounts, and that Stone keeps earning on both. What breaks it is the swipe cut and revenue-per-merchant sliding while the rate income rolls over at the same time. The pair to watch is transaction revenue, R$427mn and falling, against financial income at R$2,666mn: when the second stops rising, the first can no longer hold the total up. I am wrong if transaction revenue turns back up while financial income holds; I am right if the total is flat only because the float is doing all the work.