Pedro Conrade on building a $760M bank for the customers investors told him not to serve

The J Curve is an English-language podcast and newsletter covering Latin America's technology and venture capital ecosystem, hosted by investor Olga Maslikhova. In this episode: Nigel Morris, who co-founded Capital One inside a small Virginia bank in 1994 and grew it from roughly $1 billion to nearly $20 billion in market value, then co-founded QED Investors and backed Nubank before it existed — after David Vélez pitched him "a Capital One in Brazil" over breakfast in Mexico City. He explains why Nubank, Revolut, Klarna and Monzo are all now aiming at the same market, why banks structurally cannot innovate, why a company growing 40% with an NPS of 89 currently cannot raise growth equity, and what he thinks genuinely changed in Latin America — which is not capital, regulation, or technology.

What does a bikini shop teach you about running a bank?

Conrade opened one at sixteen, near the beach, intending to build hundreds of them across Brazil. Two lessons survived. The first was the clarity of the feedback loop: someone walks in and buys, or walks out with nothing. Success and failure were legible in a way they never are again at scale.

The second arrived when he won a scholarship to business school in São Paulo and moved away. The shop stopped performing.

“You have to be close to the business. Especially at that stage, you have to be there every single day."

Pedro Conrade

Which is founder mode, learned in a swimwear shop.

Why start a bank if you have no idea how to build one?

He is unusually direct about this: "I had no clue how difficult this was, and that's the only reason why I started. If I knew how difficult it would be, I would never have started." He cites Martín Escobari's line that entrepreneurs are slightly crazy people and agrees he qualified.

The starting point was personal irritation rather than market analysis. He was a student in São Paulo on a scholarship, with a total income of $300 a month. He didn't use a credit card, and wanted an app connected to a debit card. His benchmark was Simple, the US neobank, which paired a debit card with a personal finance manager so you could see where the money went. Nobody in Brazil had built a lighter-weight digital account — prepaid cards existed, checking accounts existed, but not the thing in between.

The cold-outreach story is the part worth stealing. Before starting, he cold-messaged the BBVA executive who had acquired Simple, on LinkedIn, asking how it worked. He got fifteen minutes. BBVA is now Neon's largest shareholder at around 30%. He did the same to reach General Atlantic — LinkedIn, cold, to a partner who had attended his school, asking for $300,000. He was told the cheque would need to be considerably larger. General Atlantic went on to lead both the $100 million Series B and the $300 million Series C.

"Early days I had no connections at all, so I had to do it. It was my only alternative.”

Pedro Conrade

Why are Brazilian banks so profitable, and what did that leave open?

Three structural conditions, in Conrade's account. First, a central bank he describes as one of the most forward-looking in the world — without the payment institution regulation and the subsequent competition rules, none of the Brazilian fintechs could have existed at all. Second, scale: the standard objection is that the US is a hundred times bigger, which is true and beside the point.

"If you have 5% of Brazilian retail banking, you're a multi-billion-dollar company."

Pedro Conrade

Third, and least obvious: Brazilians are heavy consumers of consumer software. One of the largest markets in the world for Uber, Netflix and WhatsApp, which made acquisition cheap — and also unforgiving. Early on he was asked how people would ever trust their money to a company with no balance sheet to inspect. His answer was that nobody would inspect anything; they would download the app, try it, and delete it if it failed.

That produces a genuinely different institution: regulated and balance-sheet-bound like a bank, but judged like a consumer app. His stated benchmark is not a sophisticated US wealth product. It is a big-button interface where you understand what you're doing in two or three clicks.

"Bringing banking closer to the experience of using Google, Uber, Netflix, Airbnb — that is the goal."

Pedro Conrade

Why build for the customers everyone else avoids?

He rejects the premise that it was a strategy. He was twenty-three, frustrated as a customer, and wanted a bank account that wasn't with an incumbent. The segment followed from his own circumstances — he was making $300 a month and found it absurdly expensive to hold an account. Only afterwards did it become clear the problem was widespread.

Ninety per cent of Neon's customers are now income classes C, D and E, and roughly two-thirds of the active base is self-employed with no formal salary. Investors told him consistently that this segment was hard to make money from, which he treats as the argument in favour:

"If it's hard, there's opportunity."

Pedro Conrade

There was also a sequencing benefit he didn't plan and now emphasises. Starting with a simple free account meant years of observed behaviour before Neon ever extended credit — millions of users and billions of data points already in hand when the credit card and personal loan products arrived.

How do you make money on customers who don't generate much revenue?

By competing on the other side of the equation. Neon is, on Conrade's numbers, the lowest-cost provider in Brazil — 30% below the second cheapest. If you spend materially less to serve a customer, you don't need to extract much from each one to hold a decent margin.

He credits the idea to Sławomir Lachowski, founder of mBank in Poland, who told him that if you push far enough down on cost, it becomes very hard for anyone to catch up — at which point cost stops being an operating metric and becomes a competitive moat. Conrade's summary: be focused not only on how much you make per client, but on how much you spend to serve them.

The other half is headcount discipline, and the numbers are stark. At one point Neon had 2,400 employees against roughly $200 million of run-rate revenue. In December 2025 it was $760 million with 900 people. The company is remote-first — only 35% of staff live in the city of São Paulo, nearly half live outside the state, and some of the C-level are in the US. His explanation is that a smaller, more senior team costs more per head and produces far more.

Where is AI actually working inside a bank?

Conrade is notably unexcited about the transformational case and specific about the operational one. He says Neon has not yet found a use case that would transform the business, and pushes back on the idea that agents will replace the interface entirely — for many tasks a well-built screen beats a conversation.

Internally it is a different story. Collections is the flagship: agents that negotiate around the clock, autonomously offering discounts calibrated to a customer's behaviour and completing the negotiation end to end. 85% of the collections process now involves no human at all, and he says the agents outperform some of the people.

His favourite example is smaller and more revealing. Responding to a regulator's complaint used to take a specialist half an hour of reading documents and precedents. An agent now assembles the account history, checks comparable past cases and drafts the reply; the human reads and approves. Thirty minutes to two. Eventually, he expects, some categories will go out automatically.

How do you teach financial literacy to people who don't know what they earn?

You don't. This is the sharpest reversal in the episode, and it starts with Neon's own failure. Inspired by Simple, they shipped a personal finance manager so customers could budget by category. Nobody used it. Most of the base doesn't know what they make month to month — two-thirds are self-employed — so asking them to forecast transport spending is meaningless.

So they stopped teaching and started attaching to behaviour that already exists. Conrade observed that customers reliably remember three dates: rent, the credit card bill, and school fees if they have children. Everything else can slip. Neon's savings feature therefore appears at the moment you go to pay the card: you owe 500 reais, so why not put aside 50 today. Ten months later, if something goes wrong, one instalment is already covered and you don't fall into default.

"Instead of telling them it's responsible for someone like you to have a safety pocket — forget about it. Just help them. One click."

Pedro Conrade

What predicts credit risk better than a credit score?

Engagement — and Conrade says it is roughly twice as powerful. If a customer routes their income or their sales receipts through their Neon account, that signal outweighs a weaker bureau score and justifies taking more risk. Which reframes the whole product: the account isn't the loss leader before the lending business, it's the underwriting instrument.

He also insists the differentiation is not in the products, which is the answer investors never like. A credit card is a credit card and a loan is a loan; you cannot win a feature comparison spreadsheet. The differentiation is the funnel. Every onboarding path branches on what Neon already knows — score, whether a Pix key gets registered, which product fits, whether there's capacity to save — so each customer receives a different sequence aimed at maximising engagement.

The result is the most striking set of numbers in the episode, and the three of them only mean something together. Neon has opened roughly 350,000 accounts a month for three or four years — flat. Customer acquisition cost is three times higher than it was three years ago, because the market got competitive and the days of acquiring for under a real are gone. And payback, which ran 36 months on the earliest cohorts and then 24, is under six months on the last eighteen months of cohorts.

Read that sequence again. Volume is flat, customers cost three times more, and Neon still recovers the cost six times faster. None of the improvement is coming from cheaper growth. All of it is downstream — the branching funnel, the right product at the right moment, engagement instead of scores. Which is why Conrade is comfortable saying the business can grow enormously without the headcount following.

How did Neon get its first customers?

Badly, in the sense that it worked far too well. A pre-launch teaser campaign produced 5,000 accounts on day one — and the app went down, because nobody had planned for that volume. Another 3,000 followed the next day.

Early acquisition cost was under one real. For years, more than half of signups were organic, which Conrade attributes entirely to friction: at its best, the flow took four minutes from nothing to an open account with a card in Apple or Google Pay and a transfer sent — replacing a bank relationship that had taken years of suffering to acquire.

As the market got competitive, the free ride ended and paid social became necessary. But his stated first principle hasn't changed: simplicity is the acquisition strategy.

What did Pix change?

More than payments. Before Pix, transfers were expensive, capped to banking hours, and Neon was absorbing the cost to offer them free — high cost, low frequency, therefore very little data. Pix inverted all three: free, instant, around the clock, and now more significant by transaction count than credit or debit cards.

The consequence Conrade cares about is informational. Card transaction data is fragmented across the acquirer, the network and the issuer, so no single party sees the whole picture. With Pix, Neon sees who the customer is, who they paid, how much and when — and a large share of cash behaviour has moved into that visible layer too. Together with open finance, he argues, this is what lowered the information barrier for new entrants. He won't put a multiple on the growth impact, but calls it massive.

What do you do when the central bank liquidates your banking partner?

In May 2018 Neon announced the largest Series A in Brazilian fintech history. That was Thursday. On Friday morning the central bank liquidated Banco Neon, the bank-as-a-service partner behind the product. Neon Pagamentos had 70,000 users and nothing whatsoever to do with the failure — but it had lent its brand to the partner, so when the liquidation ran across Brazilian media, everyone understood it as the fintech collapsing. Nobody had heard of the B2B partner.

The central bank, he says, was unexpectedly helpful: an official confirmed Neon bore no responsibility and offered to support a transition, with a warning attached — move fast, or customer stress will kill you regardless.

That was 7am Friday. They split into teams to find a replacement partner and map the technical migration. By 5pm they had offers; Banco Votorantim took it on and worked through the weekend on integration and documentation. By Monday afternoon customers had their money back. Tuesday was a normal day.

Conrade's account of the aftermath is unusually unguarded. Two hours of sleep across the 72. After announcing on Monday he cried for half an hour, then slept for eighteen.

What he identifies as decisive is not the deal-making but the communication, improvised in the first minute. He opened a WhatsApp group with every investor and committed to hourly updates on what was happening and what came next — and recorded dozens of videos for customers across the three days. He describes the contingency plan as something that did not exist beforehand and was built while being executed.

"I think this is basically the number one thing that helped us get where we are today.”

Pedro Conrade, on how the crisis was communicated

What did five acquisitions teach him?

That two or three of them shouldn't have happened — not because the companies were bad, but because Neon wasn't ready. "We were barely able to take care of our own problems. Imagine integrating a company." He is now, in his phrase, less romantic about M&A than he used to be, and defaults to building.

Where he thinks acquisition genuinely wins is distribution that cannot be compressed. If reaching a feature requires a network of ten thousand points of sale that would take four or five years to assemble, and buying it puts the product in front of nine million users within six months, the deal can pay for itself on that alone. What he now scrutinises first is the target's engineering team, on the theory that a weak one turns integration into a two- or three-year drag.

And the hardest part of integration, he says without hesitation, is ego — not bad ego, but a founder who ran every decision and now has to comply with someone else's process. His rule for the acquiring CEO is deliberately blunt: work out what made the company worth buying, and

“Don't kill it.”

Pedro Conrade

His method is to leave the business separate for a while, integrate back office first, run it almost as a commercial partnership, and fully absorb it only later — with the incoming CEO's incentives aligned carefully, because they are entrepreneurs too.

Why take strategic money when everyone tells you not to?

Conrade did the opposite of the standard advice, deliberately. The conventional line during the boom was that strategics on a cap table are dangerous. His counter is specific to banking: this is a cyclical industry, and Latin America produces a macro shock roughly every decade, so you want people who have survived a credit downturn before.

In practice that means a BBVA risk specialist sitting on Neon's risk committee and General Atlantic people covering other blind spots. And when credit deteriorates, an investor who has seen the cycle says hold your nerve for six months and trust the cohorts, rather than panicking.

Neon now has eighteen institutional investors — BBVA, General Atlantic, BlackRock, PayPal, Monashees, Quona, Propel, Vulcan, IFC and DEG among them — and close to $1 billion of equity raised. Conrade controls the board and is the controlling shareholder. His point about that is the interesting one:

Not a single decision in these ten years did I have to say this. It was always consensus. Even if it took ten meetings to get there, we always got there.

Pedro Conrade

On pitching, he is self-deprecating — he doesn't think he was ever good at it, having raised a billion dollars — and lands on the mechanism that actually worked. Nobody was going to write a large cheque to a twenty-three-year-old on a first meeting. What compounded was a track record: promise 10× growth and deliver 15; promise 5× and deliver 7. Do that a few times and the pitch stops mattering.

Why does a founder hand over the CEO job at 33?

Because he doesn't think he was ever the right person for the role — and because he had already given away most of the equity that would make him cling to it. Of the management pool, the team holds 75% and Conrade holds 25%. He describes a handful of people who joined four or five years in as late co-founders who own the business as fully as he does and sometimes make more consequential decisions.

His description of his successor, Fernando Miranda, now CEO of Neon Pagamentos, is a portrait of the discipline he says he lacks: up at 4am, marathon, breakfast with the children, in the office before eight having read everything, out at six, then drinks with the team and dinner with the family. Conrade calls him the pacekeeper — the person who converts a 100% target into a demand for 150%, which he says a company at this scale requires and he cannot personally supply.

What he keeps is product and talent. He goes through every Figma file, tests the product daily, is the most active reporter in the internal bug channel, and spends about 30% of his time on recruiting and unblocking people. His framing of the split: stepping out of the CEO role was easy; finding the right person was hard.

What does Neon look like in ten years?

Smaller than it should be, in his view. "We're still very small. We can grow ten times and still be small." Geographic expansion — other Latin American markets, possibly Southeast Asia — is acknowledged as an option but explicitly not the plan, because he believes there is a 10× still available inside Brazil. An IPO is the assumed direction.

He is also candid about why he is not optimising his life around the company. He lost his father young, which is why there was a bikini shop at sixteen, and says it taught him early what matters. His ranking is family, then business, then himself — with a note that he is trying to move himself up a place, because he is thirty-three and, in his own assessment, looks forty-five.

Rapid fire: six questions

Three things true in Latin America that aren't true in Silicon Valley. Resilience, because the cycles will always come. Bring in global talent — the local people are brilliant, but other cultures and other expertise compound. And don't underestimate what a Brazilian or Latin American business can become; it can be global.

What he'd do differently in the first 90 days. Hire fewer people and hire more senior ones. Build one thing at a time.

The most under-the-radar company in Latin America. Eve — the Brazilian eVTOL company built by Brazilian engineers, which he thinks could transform the world, with the caveat that it's already listed so may not qualify as under the radar.

The question founders should ask investors and don't. What does success look like for you, when, and why — then check whether you are actually helping them get there. He says he asked it recently for the first time.

What would surprise people who know him professionally. That the job was never first. Family was always the priority, then the business, then himself.

Where he'd build next. Basic needs — housing, health, food, finance, transport. He mentions friends who started businesses he could barely understand the need for, and his own instinct to build around the things people cannot avoid doing.

"People need to live, make money, eat. There is plenty of room for innovation, especially in Latin America, for something like this."

Pedro Conrade

COMPANIES AND INSTITUTIONS

⠀PEOPLE

⠀IDEAS AND FRAMEWORKS

Founder and executive chairman of Neon Pagamentos, the Brazilian digital bank he started at twenty-three. Opened his first business, a bikini shop, at sixteen. Neon has raised close to $1 billion from eighteen institutional investors including BBVA, General Atlantic, BlackRock and PayPal, serves around nine million customers concentrated in Brazil's lower and middle income classes, and reported roughly $760 million of run-rate revenue in December 2025. He stepped back from the CEO role and remains the controlling shareholder.

Nigel Morris, QED — Capital One, Nubank and what actually changed in Latin America · Stelleo Tolda — How MercadoLibre beat Amazon in Brazil · Brian Requarth, Latitud — why ~50% of LatAm founders now go global · Hernán Kazah, Kaszek — building Latin America's $100B+ companies

🎙 The J Curve  is where LATAM's boldest founders & investors come to talk real strategy, opportunity and leadership.