The Reserve Jump Inside Nubank's Blowout Quarter

Nubank set aside $1.79 billion for potential credit losses in Q1 — up 33% from the previous quarter on a currency-neutral basis. That's a big jump in a short time.

The concern isn't just the size of the reserve. It's the context. Nubank's total credit portfolio grew 40% year-over-year to $37.2 billion. Loans more than 90 days past due — the ones most likely to default — sat at 6.5% of the portfolio.

Earlier-stage delinquencies (15 to 90 days) were at 5.0%. That pipeline of souring loans is what matters most — it tells you what the 90+ number will look like next quarter.

When a lender is growing its loan book that fast and simultaneously setting aside a third more money for losses, the market reads it as a warning sign. The question investors are asking: is Nubank growing because it's winning, or because it's lending to people who can't pay back?

Nubank's credit card portfolio was $24.3 billion in Q1. Its unsecured lending book was nearly $10 billion. That means roughly $34 billion of its $37 billion loan book has no collateral backing it.

Nubank's customer base in Brazil alone surpassed 115 million people — in a country of 215 million. That kind of penetration means Nubank isn't just serving the affluent. It's deep into lower-income segments where household budgets are tighter and more sensitive to economic shocks.

This is the domino that worries credit analysts most. If Brazil's economy slows or interest rates stay elevated, these borrowers feel it first — and there's no collateral cushion to absorb the blow.

Nubank earns in Brazilian reais but reports to Wall Street in U.S. Dollars. When the dollar strengthens — as it did this week amid rising U.S. Bond yields and inflation fears — the same amount of reais translates into fewer dollars. Profits shrink on paper even if nothing changed on the ground.

But here's the non-obvious part: Nubank's loan loss reserves are also built in reais. When those reserves get converted into a stronger dollar, the cushion protecting against bad loans looks healthier than it really is in the U.S. Filing. A 249% coverage ratio reported in USD could mask a thinner real-terms buffer if the real is weakening.

So investors reading the English-language filing may see a rosier picture of credit protection than what actually exists. The loans are made — and go bad — in local currency. The FX translation doesn't just compress earnings — it can flatter the safety metrics at the exact moment they matter most.

Nubank's management described its U.S. Expansion thesis as "bounded downside, uncapped upside". They've capped the investment at less than 100 basis points (one-hundredth of a percentage point) of impact on the company's efficiency ratio in both 2026 and 2027. That sounds disciplined.

The deeper concern is about credit modeling. Nubank built its lending models on Brazilian data. That means spending habits, income swings, and default patterns unique to Brazil's economy. In the U.S., Nubank is targeting people with thin credit files and limited bank access — a group that looks a lot like its early Brazilian customers on the surface. But porting credit models trained on one country's data to underwrite loans in another is where fintech lenders have historically gotten burned.

Nubank's efficiency ratio was an impressive 17.6% in Q1 — extremely lean for a bank. That gives it room to absorb early U.S. Losses. But picture this: the models get American defaults wrong while Brazilian loans are also going bad. Nubank would be fighting on two fronts with one pool of cash to cover losses.

Nubank's Q1 return on equity was 29%. Revenue grew 42%. As of May 16, 2026, the stock was trading near its 52-week low of $11.71, down roughly 21% since late April.

On top of that, Nubank's coverage ratio — the amount of reserves it holds relative to its worst loans — was 249% in Q1. That means for every dollar of loans 90+ days past due, Nubank has $2.49 set aside. That's a thick cushion.

The bull case is straightforward: the market is pricing in a credit blowup that may never come. The bear case is just as clear: Nubank is lending to the riskiest borrowers in Latin America. The 33% jump in money set aside for bad loans? That's the first tremor before a bigger quake. The next two quarters of credit data will settle the debate.