PagSeguro Digital Ltd. PAGS

Key data

The business

PagSeguro sells a Moderninha card reader to Brazil's long tail of micro-merchants and small businesses, then wraps them and their customers in a digital bank, PagBank, that holds deposits, moves money over Pix, pays bills, and increasingly lends. It makes money three ways: a take rate on the R$133 billion of card volume it processes a quarter, fees and float on 34 million banking accounts, and a spread on a R$5.1 billion credit book. The moat is distribution plus stickiness: once a merchant's money, Pix, and working-capital loan all live in one app, leaving is expensive. The company even swept dormant accounts this year, charging a residual fee and dropping them from the active count, a tell that it now manages the base like a bank, not a growth-at-any-cost acquirer.

The numbers

The shape is a maturing core and a compounding bank grafted onto it. Revenue nearly tripled from 2020 to 2024, but the annual pace has cooled to the low teens as card acquiring saturates.

Revenue and net income for 2026 year-to-date were not available in this run's structured filings. Margin fell as the mix shifted from high-take acquiring toward lower-margin volume, then stabilized near 11% as banking earnings filled in. Quarterly earnings have been steady and slightly ahead of what analysts modeled, the Q2 2026 print of $0.41 the fifth straight quarter around $0.40.

The engine underneath tells you where the growth actually is.

Card volume grew 3% while revenue excluding interchange grew 1.7%, which means the take rate on acquiring is still compressing under Cielo, Stone, and Mercado Pago. But cash-in rose 23% and the loan book 31%. If credit compounds near 30% and deposits keep funding it at 0.9x book, earnings can grow double digits while the card business flatlines. At 6.4x earnings, the market is not paying for that arithmetic. What it believes instead is either a permanent stall or a credit book that seasons badly, and the print that settles it is the delinquency line as the loans age.

Management

Insiders disclosed in Form 4 filings show Luis Frias bought about $5.0 million of stock in two lots on March 27, 2026, while Alexandre Magnani sold about $2.25 million on April 17; across twelve months insiders bought $5.0 million against $3.4 million sold, with plan status not disclosed on the sales. Frias adding at 6x earnings is the signal that carries weight; officer sales are the smaller, noisier counterweight. Capital return backs it: the stock trades below book at 0.93x, so buybacks are accretive, and return on equity sits at 14.6%. Pay and guidance detail were not in this run's structured filings.

How it fails or surprises you

Credit quality (downside). The loan book grew 31% to R$5.1 billion, pushed into working-capital and unsecured lending where losses show up late. Under Brazil's interest-rate environment, a bad vintage could turn provisions into an earnings sink. Delinquency detail was not disclosed this run, so the Q3 2026 credit lines are the first honest tell, and this is the fact the case for owning it explains least.

Acquiring erosion (downside). Volume up 3% and ex-interchange revenue up 1.7% say the card business is maturing into a price war. If acquiring revenue turns negative before banking fully scales, the whole machine stalls at once. Watch net take rate and volume together, not either alone.

Re-rating (right tail). At 6.4x earnings and 0.9x book, near the lowest since 2018 against peers at 24x, any proof the bank pivot works, deposits and credit compounding while losses stay contained, could re-rate the stock toward its own 9-to-35x history. Brazilian monetary policy has been easing from cycle highs, so the tailwind is intact but not yet fully extended.

Closing thoughts

This is a name already priced for melting payments, so the edge is not in spotting the decline, which everyone sees, but in weighing the bank growing inside it that the 6x multiple ignores. On the other side of the table sits a market that has watched Brazilian fintechs lean on a growing loan book to outrun a tired core, then meet the vintage that seasoned badly. The settling print is credit quality over the next two or three quarters. The tail looks fatter to the upside because 6x earnings and sub-book value are real downside protection, but that protection evaporates if the loan book cracks, so the whole distribution hangs on one metric the company did not disclose this quarter.

The bet is still that Brazilians keep banking on their phones with PagBank, keep more money parked there, and borrow more of it, while the card machines PagSeguro rents to Brazil's small shops fade slowly enough for the banking side to carry the company. What breaks it is credit going bad, or acquiring collapsing before banking is big enough to catch it. The one pair that tells you first is the credit portfolio's growth set against its delinquency rate, and until the delinquency half of that pair is on the page, the read is unproven.