What Brazil's free rail killed, what it gave away and where the margin went. The lesson for MODO, Mercado Pago, Ualá, Pomelo and ueno
TL;DR
In November 2020 Brazil's Central Bank (BCB) launched PIX, an instant payment rail that is free for people and mandatory for large banks. The reading that has circulated ever since says it ate the private wallets' market. The BCB's own open data tells a different story, and a more useful one for anyone building payments in the region.
What PIX killed were instruments: DOC (the slow, cheap interbank transfer) went from 250 million transactions to exactly zero, TED (the immediate, expensive one) lost 57% of its operations, cash withdrawals fell 36% and debit cards had their first annual decline in 2025. The wallets, meanwhile, grew riding on the rail: Nubank, PicPay and Mercado Pago now hold 36.6% of all PIX keys, more than Itaú, Bradesco and Santander combined. And in September 2025, a month I computed on the BCB's granular dataset and that is published nowhere, the most important thing happened: for the first time Brazilians used PIX more to pay merchants than to send money to each other.
The thesis of this case: PIX gave the wallets the rail and confiscated their moat. The transfer layer became free by mandate, so the system's margin moved to the one place the rail does not touch: credit. The winners were those who could lend, the losers were the owners of the old rails, and the whole market grew wider at the bottom. At the end, what this means for MODO, Mercado Pago, Ualá, Pomelo and ueno. Independent analysis, not affiliated with any of the companies mentioned.
PIX won on three design decisions no private player could make, rather than on the quality of its app. They are documented by the BIS, the body that groups the world's central banks, in its report on the system: participation is mandatory for large institutions, so the network was born complete instead of growing in fragments; usage is free for individuals by mandate, so nobody can undercut it on price; and the brand is single and owned by the regulator, so the butcher and the bank display the same logo.
There is a fourth decision that usually falls out of the story and sets the board: in June 2020, five months before launching PIX, the BCB suspended WhatsApp Pay, which had launched with a 3.99% merchant fee. The regulatory message was clear: Brazil was not going to have a private closed loop dominating person-to-person payments. That fee pool stayed empty by regulation before anyone could fill it.
To size what came next: PIX went from 176 million transactions in its first two months to 79.6 billion in 2025 (452x), moving R$ 35.3 trillion, according to my calculation on the BCB's payment-methods series. The BCB's own survey shows cash falling from 42% to 22% as the most used payment method in three years.
The right question is at whose expense it grew. The answer sits in the BCB's quarterly series.
The instruments whose only job was moving money died: DOC reached exactly zero and was formally switched off in January 2024, with Febraban explaining that customers preferred PIX "por ser gratuito e instantâneo". Cheques lost 62% of their operations. Cash withdrawals, 36%.
Two nuances block the simplistic story, and both matter. The boleto did not die: its transactions rose 9% and its value reached R$ 10 trillion in 2025; the instrument that bills companies survived the free rail. And TED forked instead of dying: it lost 57% of its operations while its value rose 61%, because it became the rail for large money. In June 2026 TED still moved more money than PIX (R$ 4.2 trillion against R$ 3.6 trillion in the month), with PIX doing 120 times more transactions. The new rail took the people; the old one kept the treasury.
Debit is the victim still standing. It grew 43% between 2020 and 2025, but in 2025 it had its first annual decline of the series, and the acquirers had been flagging it: for Stone, PIX "cannibalize on debit volumes... we do not see PIX cannibalizing on credit volumes"; for PagSeguro, "probably PIX is cannibalizing the growth of debit". The reason is price: a Brazilian merchant pays 0.22% on average for PIX against ~1.1% for debit and ~2.2% for credit. Against debit, PIX competes head to head and cheaper. Credit it barely touches, because credit sells financing, which the rail does not include.
In its first December, PIX was a transfer app: 84.8% of operations were between people (P2P) and only 6.4% went to merchants (P2B). Five years later, in December 2025, merchant payments were already 44.4% (3.15 billion transactions in the month) against 42.4% for P2P. The crossing happened in September 2025, by a thin margin I will state: 677,000 transactions out of 6.13 billion, one hundredth of a point. From there the gap only widened: two points by December, almost seven by July 2026. This is my calculation on the BCB's granular dataset; I found no publication that had done it.
That crossing is what reorders the industry. While PIX was P2P, it competed against cash and DOC. Once it became a checkout network, it started competing against acquiring, which was the business of Stone, PagSeguro and Cielo: Bloomberg documented Stone and PagSeguro falling from some USD 50 billion in combined valuation to about a tenth of that. One nuance: not all of that collapse is PIX (Cielo, for one, lost share to Stone and PagSeguro themselves before being delisted), and Stone today reports record average fees while integrating PIX QR into its terminals. The cheap rail hit valuations and at the same time became an input for those same companies.
In the BCB's key registry, my calculation shows PicPay going from 6.0% of the stock in December 2021 to 10.1% in July 2026 (already above Itaú back then), and Mercado Pago from 4.8% to 7.7%. Adding Nubank (18.8%), the three wallets now hold 36.6% of the system's keys, more than double Itaú, Bradesco and Santander combined. The business numbers follow: Mercado Libre reported 78 million fintech users and USD 278 billion in total payment volume (TPV, +41%) in 2025, and PicPay reached Nasdaq after doubling its profit. PicPay's prospectus puts it in writing: PIX "has been favorable to our platform". And Giménez, of Mercado Pago, had already described it back in 2021 as "incremental volume".
So the rail gave them everything? Not quite, and here is the price they paid:
PIX gave the wallets the rail and confiscated their moat. Moving money became public infrastructure: P2P funded with balance is free by mandate, the closed loop was banned in practice, and the system's margin moved to the one place the rail does not touch: credit.
PicPay's prospectus documents it with accounting precision: PIX transfers funded with balance are "free of charge"; the company monetizes when the funding source is a credit card (3.49% to 5.49%) and through Pix Credit, which moved R$ 19.6 billion in 2024 (+69%). 52% of its 2025 revenue already comes from credit, and the company now describes itself as one of Brazil's leading digital banks. Nubank turned PIX into an interface and a credit originator. And the most telling signal of who kept what: the BCB dropped its plan to regulate an official Pix Parcelado and left the financing to private hands, at rates around 5% monthly. The State kept the rail; the margin stayed upstairs, for whoever holds a licence and a balance sheet.
Argentina already shows the same pattern of victims. The retail payments report from the BCRA, Argentina's central bank, for July 2025 records debit falling 31.5% year on year in counts and ATM withdrawals 40.5%, with credit rising 12.9%. The Brazilian sequence, delayed and without a PIX: here the rail is Transferencias 3.0 and the interoperable QR. What each type of player can read from the case:
| Claim | Status |
|---|---|
| PIX 2020→2025: 176 M → 79.6 B transactions; R$ 35.3 trillion | My calculation on BCB open data |
| DOC at zero; TED −57% count / +61% value; cheque −62%; withdrawals −36% | My calculation on BCB open data |
| Debit: first annual decline in 2025 | My calculation on BCB open data |
| P2P→P2B crossover in September 2025 | My calculation (BCB granular dataset); unpublished |
| Keys: Nubank 18.8%, PicPay 10.1%, Mercado Pago 7.7% (Jul 2026) | My calculation on the BCB's DICT |
| PIX costs merchants 0.22% vs ~1.1% debit and ~2.2% credit | BIS (minor inconsistency between its papers) |
| WhatsApp Pay block; DOC discontinued; Parcelado deregulated | Published facts with sources |
| Quotes from Stone, PagSeguro, PicPay and Mercado Pago | Filings with the SEC (the US securities regulator) and public transcripts |
| Stone+PagSeguro valuation falling to ~a tenth | Bloomberg (Nov 2023); partially attributable to PIX |
| Argentine echo (debit −31.5%, ATM −40.5%, Mercado Pago 82% of balances) | BCRA, retail payments report, July 2025 |
| Regional echo: Ualá pricing, SPI 97.2%, ueno 2.7 M customers | Published with sources |
| MODO: 7.5 M active users | Self-reported by its CEO, unaudited |
| "The margin moved to credit" | My interpretation on top of the data above |
The status column governs how each number reads. The own calculations are reproducible: the case scripts download the BCB datasets and regenerate every figure and all three charts.
The BCB does not publish PIX transaction volume per participant, only keys. If that data appeared and showed wallets losing transaction share to the banks, the thesis would move towards the reading this case discards; that hole is declared.
If the wallets' growth turned out to be attributable only to general digitalization and the pandemic, the "rail as a gift" would be overstated: the counterfactual of how much they would have grown without PIX does not exist, and no accessible causal study resolves it. And if the BCB ever charged for the rail or regulated the credit margin on top of PIX, the regional conclusion (the margin stays upstairs, with the private players) would need rewriting: the December 2025 signal points the other way, but it is a regulatory decision and it can change.
analysis/pix-brasil.py reproduces every own calculation and charts/pix-brasil.py regenerates the three charts; the BCB dataset URLs sit in each script's header.If you read this far, something about the problem caught your interest. I like discussing these decisions with people who live them: write me what you would do differently, or let's book a call.
See all projects