In case you missed it:

Read my Nu Deep Dive here to learn about the company in detail.

Brazilian 15-90 day non-performing loan (NPL) improvements were slightly beyond expected seasonality, while the 90+ day NPL number was exactly in line with seasonal patterns. They still don’t disclose Mexican credit quality, but they said it’s performing in line or “better than expected” in some areas. One of the analysts in the Q&A who tracks Mexican NPL data said the trend is improving as part of his question to the team. Nu didn't push back.

Nu’s team was asked several times about why provisions looked so good in relation to portfolio growth. How are they delivering 14% FXN Q/Q growth in originations AND a 7% FXN Q/Q decline in credit provisions? Do they think this is a one-off or is it the new norm? Encouragingly, they think a lot of it is structural in nature.

First, their credit portfolio is performing better than expected. As we’ll dig into later, AI investments and ongoing underwriting improvements informed by a perpetually growing base of data are helping. They feel these improvements are just beginning. Next, they are improving overall risk by intentionally moving to higher-quality unsecured borrowers, including more business loans. They’re also augmenting credit quality by leaning aggressively into secured lending, which will be great for credit book resilience amid weak parts of cyclicals. Recall that secured lending growth is outperforming both credit cards and unsecured.

This rapid credit growth is not being driven by loosening their risk parameters and originating to riskier customers. Quite the opposite. Beyond these two items, there was a temporary benefit that also helped. This part isn't structural. During the quarter, they offered credit access reactivations for defaulted customers who paid down all outstanding debt. This led to a big quarter for recoveries, which helped credit loss allowance expense during this specific quarter.

Moving onto NIM, the aforementioned intentional shift to prime borrowers is also weighing on this metric. Fortunately, while NIM fell Q/Q, risk-adjusted NIM rose Q/Q by 70bps. That shows the reduction in interest income is being more than offset by the reduction in provisions thanks to lower risk. It shows this is worth it. Nu leadership sounds confident that this is a maintainable theme and not reliant on the aforementioned Q3 recoveries spike.

“As we have done in the past, if and when we see any deterioration in asset quality across any of the segments, any of the products, or any of the geographies, we will not hesitate to kind of pull the brakes, reassess and revisit where to go.” – Co-Founder/CEO David Vélez

Nu trades for 21x forward EPS. EPS is expected to compound at a 38% clip during 2026 and 2027. Estimates should be relatively stable following this report.

It doesn't offer guidance.

Thriving in Brazil:

Nu added ~3M net new customers in Brazil, which compares to ~3M last quarter, ~2.5M 2 Qs ago and ~3M Y/Y. 11% Y/Y customer growth for a company that already has 60% of the adult population is quite notable. And despite these rapid additions, active customer rate in that nation remains at 85% and well better than competitors.

Thriving in Mexico:

Since launching in Mexico about 6 years ago. It already has 14% of the adult population as customers, which is better than the 10% penetration rate it had in Brazil at a similar stage of maturity. In just two years, they’ve gone from 4M customers to 9M and now to 13M. On an apples-to-apples time-in-market basis, Mexico has 3x more deposits and 2x more ARPAC than Brazil. Mexico’s $12.50 ARPAC is actually already approaching Brazil’s $13.50, despite just recently launching traditional loan products that it already offers in Brazil. Everywhere you look, there’s reason for strong optimism in this market. The ARPAC ceiling looks a lot higher than Brazil. The population has 40% higher income per capita than Brazil... despite having lower credit card penetration rates. Mexico also has a much higher revolving credit mix than Brazil, which enables better unit economics and returns.

Despite all of the positive momentum in Mexico, deposits fell Q/Q from $6.7B to $6.1B. This was related to their decision to shift some of the focus away from customer growth and towards margins now that they have built scaled customer and deposit bases. They’re still mainly prioritizing customer growth in Mexico, just not as intensely or singularly as before. Nu's brand awareness and product have gotten to a point of maturity where they don’t need to pay sky-high deposit yields to attract customers and deposits. They no longer need to operate at poor margins to grow rapidly. While there was a short-term dip in deposits as a result of this change, that churn reaction was in line with their expectations. And furthermore, recent trends make them confident in more growth going forward. This was a one-off and sets the company up for much more profitable and sustainable growth ahead.

A Broadening Durable Competitive Edge:

Nu’s app is fantastic, its brand is ubiquitous and its team is highly capable. Those items all make me confident in its future. At the end of the day, however, this is still mainly a bank. We talk about it constantly. Banking is a commodity. Relative cost advantages are the way to profitably pass on more customer value and delight to stand out from a crowded pack. Nu has had a large monthly cost to serve advantage in Brazil for years. Incumbents are routinely at or above $5 per month and Nu has ranged between $0.80 and $1. That’s how they undercut competitors on loan interest rates, how they offer more deposit yield, how they add more perks to their loyalty programs, how they fund future innovation and how they win. With that in mind, it’s so encouraging to see Mexico looking like Brazil. They’ve moved from a $3 monthly cost to serve in 2021 to $1 today. I couldn’t find a reliable estimate of Mexican incumbent bank cost to serve, but the wide range of estimates were all factors higher than $1.

They've matched their Brazilian edge in Mexico.

“We think that in Mexico, we enjoy a very favorable cost structure compared to many of the other players in the region. That allows Nubank to play in some segments that incumbent banks are unable or unwilling to play in. They can’t price it lower and still have compelling unit economics.” – Co-Founder/CEO David Vélez

Developing in Colombia:

Nu’s Colombia business is now up to 4M customers or 11% of that adult population. It continues to add deposits at a solid clip, with $200M added this quarter vs. $300M Q/Q and $900M last year as it was ramping its nascent high-yield offering in that nation.

Existing Customer Runway:

As we start to see signs of Nu exhausting its Brazil customer growth opportunity, ARPAC will become more important. 11% customer growth is the slowest I’ve seen them record and that’s entirely unsurprising. They cannot add millions of Brazilians every single quarter. They already have 6 out of 10 adults in their ecosystem. They will run out of people to add and they won't own 100% of the market, as Inter & Co, Itaú and Mercado Libre are not going anywhere. That’s why its ARPAC opportunity is so compelling.

That figure has compounded at a 17% clip over the last 2 years to reach $13.50 in Brazil. As of last quarter, their most mature cohort had a $27+ ARPAC… which has been steadily rising. Incumbents have a $40 ARPAC because more of their product suite is developed. Nu’s product suite is getting there and its ARPAC should easily approach $40 as that happens. This will be a fantastic Brazilian growth amplifier as they keep rapidly expanding in their other 2 markets. Here’s another way of putting it. Their primary bank account market share is 6x higher than their gross profit market share in Brazil (30% vs. 5%). They have such a large opportunity to cross-sell more and more products and approach $40 down the road (with very little marketing spend to get there). Considering its push into commerce, telecom and more, you could easily argue their ceiling is higher than $40.

Credit Portfolio:

Credit card balance growth accelerated as Nu significantly raised limits in Brazil. This was mainly for high-credit-quality customers and has already helped accelerate purchase volume. That boost will not be a one-quarter benefit, considering Nu is rolling out these increases in phases. They expect the full benefit to be felt towards the end of next year, so the growth accelerant should be gradual and steady for a while. And looking ahead, they see plenty more opportunity to responsibly raise limits even more.

Secured lending rose by 133% Y/Y, with its public payroll loan product powering the growth alongside a recovery in Brazilian Institute of Social Security (INSS) volume. As a reminder, fraud within the INSS program was uncovered and led to a temporary sharp decline in activity. That has since normalized as expected. The secured FGTS (mandated Brazilian savings account) loan product is also under regulatory threat that could materially lower the product’s demand. Nu fully expects to offset that potential headwind with demand for public payroll loan products. Great to hear.

For secured private payroll loans, they’re still not ready to lean into originations quite yet. The cost of risk is too high for the product as a whole based on its current structure. They’re waiting for changes to play out to make the unit economics attractive enough to lean in. They will not chase irresponsible growth. Lastly, unsecured lending grew by 63% Y/Y as Nu prioritizes secured lending first, unsecured second and credit card third.

AI:

Nu is determined to use the latest and greatest AI models and agents to improve every facet of its operations. It wants the app experience and product delivery to get a lot more personal, with agentic experiences uplifting overall value. Internally, it aims to use this technology to boost collections (like it did this quarter), cut fraud losses, cut provisions (again like it did this quarter) and streamline all other cost buckets. If done well, this will diminish Nu’s cost to serve and boost its efficiency even more, enabling deeper differentiation amid a competitive sector. They’ve already developed “Nuformer,” as a structured approach for training and deploying company-specific models, with initial deployments directly leading to the successful Brazil credit limit boosts already discussed. They have many more models planned and so much more data to infuse into them. And encouragingly, they’ve only applied this work to existing customers in Brazil. Not new customers and not in their other two markets either … which is all coming.

Other Notes:

Very good quarter. It’s impressive to see them already using AI to improve various core parts of their business and refreshing to hear a bank (aside from SoFi) fixating on this disruptive technological revolution. Demand trends are fantastic in Brazil, with soaring engagement levels greatly extending the growth runway beyond finding new customers. Mexico has already turned into another home run for them and Colombia momentum is respectable as well. Their growth engine is humming and their margins remain quite resilient despite aggressive expansion into new markets. They could be making a lot more money right now if they wanted to, but that would be short-sighted. It's about maximizing profit dollars over the decades, not quarters, to come. That entails investing heavily today.

It’s hard to find fundamentals that are this impressive for a name trading at 20x forward EPS. I understand that Nu comes with considerable Latin American credit risk and that deserves some level of discount. This needs to be deeply respected and watched closely (on it). At the same time, the 0.5x PEG ratio is already a big discount. And furthermore, the credit book is getting more resilient, their performance has been strong across a wide array of macro backdrops in Brazil and the team is responsible. They preemptively pull back as soon as they see the slightest sign of anything going wrong, and I find that comforting.

I remain a confident shareholder.