Nu added 3.7M customers for a second consecutive quarter to reach 142.6M. This represents 12% growth and marks ongoing maturation for this part of the growth curve. That’s what happens when 65% of Brazilian adults are already on your platform. Going forward, more of the growth will come from other countries as well as engagement gains from their existing customer base. And while they have dominant customer market share, their gross profit market share remains in the low-to-mid single digits depending on the product and their average revenue per user is under $18 compared to over $30 (and quickly rising) for their most mature cohorts. With active customer rate rising to 83.6% and at best-in-class levels, they have a massive base of engaged consumers clamoring for more.

On deposits, 19% FXN growth was again held back by intentional moves in Mexico to cut their funding costs, accept some churn and gear their unit economics towards something poised for a lot of credit book scaling.

Within 33% Y/Y FXN credit portfolio growth, credit cards rose by 32%, unsecured rose by 39% and secured rose by 23%. As a reminder, FGTS is a mandatory savings program in Brazil. Last year, the government tightened borrowing rules and requirements around obtaining these loans, which led to 80% shrinkage in volume. This was industry-wide, and obviously hit NU too considering its presence there before the change. They’ve offset that hit with a lot more public and private sector payroll loans, where they’ve gained much more confidence (especially on the private side) in risk pricing and market dynamics, allowing them to accelerate originations.

As a reminder, NU reports “managerial” (adjusted) revenue alongside the typical revenue number. This adds Brazil revenue taxes back to the top line to show the impact it has on them. At the same time, there’s no difference for net income, as that same cost is just incurred further down the income statement.

A one-time counterparty event weighed on gross profit margin (GPM) this quarter. Excluding it, GPM would have been about 43.0% vs. 41.9% reported. The charge was due to a batch of receivables it bought from institutions that the Brazilian Central Bank forced them to offload. This cut risk-adjusted net interest margin (NIM) by 63 bps. This will not recur and has nothing to do with loan performance. The Brazilian government is cracking down on fraud from companies like Entrepay and MW Pay. Nu had some of their receivables on its balance sheet because these firms sit between the card issuers (Nu) and the merchants. A collapse can lead to owed money being tied up and eventually lost, and Nu was already suing for fraud because of this. Again, without this one-time charge, GPM would have reached its highest level in a year. Furthermore, there was another 55 bps hit from Brazilian debt renegotiation programs, making the GPM result a lot more impressive. 

OpEx was a bit lighter than expected. Nu hired fewer people than it thought it would due to AI-driven advancements while some other costs slipped from Q3 to Q4. This helped the efficiency ratio improve from 20.3% to 19.3%. As a reminder, efficiency ratio is simply OpEx/revenue, with lower being better. For context, Itaú, Brazil’s largest private bank, sits at 37.5% despite a far larger scale and longer history.

Definitions:

Nu has $5.7B in cash & equivalents (including holding co.) on top of its $4.2B capital requirement. Credit loss allowance did rise from $6.6B to $7.1B Q/Q, but that was mostly due to rapid ongoing credit book growth. Growth drove $332M of the $500M rise, with $135M coming from intentional risk expansion. Risk-adjusted net interest margin (NIM) (NIM after provisions) is a great way to see whether or not they’re getting properly paid for this added risk. As you can see below, they clearly are.

On NPL deterioration, this is also orderly and as expected. They are accepting wider customer types because the added interest they’re getting is more than making up for it.

It follows an intentional step-up in risk-taking starting during Q1 2026 and they haven’t slowed down since. While others like Sea Limited talk about credit issues and needing to pull back in Brazil, Nu remains aggressive. Again, strong risk-adjusted NIM says they’re doing very well.

Credit Quality:

The 15–90-day non-performing loan (NPL) rate rose to 4.9% from 4.8% Q/Q, and the 90+ day rate rose to 7.3% from 6.9%. Seasonality lowered the 15–90-day rate by 9 bps, while intentional risk expansion added 13 bps. Leadership calls that a deliberate trade-off. It’s confident that expected returns justify the extra risk given the strength of its underwriting and how resilient its credit performance has been. And there’s solid evidence of them being right. The same things that hold back GPM hold back risk-adjusted NIM, meaning normalized risk-adjusted NIM would have been 12.7%, marking another 30 bps of sequential improvement.

Beyond that, when looking at NPL as a percentage of interest-earning balances rather than the overall book, the 15-90 day NPL rate improved from 8.2% to 8.0%. This excludes the mix-shift impact as this type of credit moved from 59% of total to 61%. It’s direct evidence that added interest income more than offsets the formulaic rise in losses.

Why is Credit so Resilient?

So why do things still look so good ex-mix shift and how is it continuing to accelerate originations in Brazil while others pull back? A few compelling reasons. The first is the most telling and gets at their culture and approach to building a business. Macro has steadily gotten worse in Brazil since they started scaling 13 years ago. So? They just assume it’s going to keep getting worse going forward and only underwrite loans that will work in that outcome. That especially matters right now, with Brazil’s Central Bank saying 20.3M people carry risky debt levels as of June, up 75% in five years.

This absolutely caps some origination growth during the rare fun times in Brazilian macro, but it also guards their balance sheet and is partially why they can keep growing so quickly during the not-so-fun times. Next, they’re still very small in terms of credit market share (again much bigger for customer market share). There’s a ton of low-hanging fruit for them to pick the highest-quality customers from their competitors, so that’s where the initial focus has been. Quality begets resilience, while market share gains always have a way of buffering any macro-related pain. Next, their primary banking relationship penetration is over 60% in Brazil. This gives them access to a much wider array of asset and repayment data to sharpen underwriting per borrower. And finally, they do think they’re best-in-class at underwriting. They think their AI-native DNA and core helps them use information more effectively than others can and price risk superiorly. That last argument is highly subjective, but the others are quite concrete. All of this is why the industry continues to see some degradation while NU doesn’t.