A-Portfolio

The company in one breath

Nu Holdings, ticker NU on the New York Stock Exchange, runs Nubank, a branchless financial platform serving consumers and small businesses across Brazil, Mexico, and Colombia. Customers use one mobile app for deposits, cards, personal loans, payments, investments, insurance, and other daily money jobs. Nu wins when cheap digital service, richer customer data, and broader product use let it earn more from each customer without importing the expense base of an old bank.

This is Latin America's largest digital bank, with 138.9 million customers, $1.06 billion of quarterly net income, and a 33% return on equity. The question is how much perfection belongs in the price.

Why now

The latest verified regular-session trade was $14.55 at 3:59:59 p.m. Eastern on August 31, from Alpaca's IEX broker feed. The stock closed 1.8% above its prior-session close of $14.29, but remained 23.3% below its 52-week high of $18.98. That pullback gives us a cleaner look at a business whose operating numbers keep getting stronger.

Nu's August 13 report was a monster. Currency-neutral gross revenue rose 39% to $5.88 billion, gross profit rose 43% to $2.44 billion, and net income rose 49% to $1.06 billion. Customer growth slowed to a still-healthy 13%, while monthly revenue per active customer rose 22% to $17.10. That is exactly how a maturing platform should evolve: fewer points from adding names, more points from becoming the main financial relationship.

Mexico supplied the fresh catalyst. Nu began operating as a full bank there on August 6 after reaching 16 million customers in July. That license expands the products and payment capabilities it can offer. Mexican customer groups also reached monthly revenue per active customer of $12.30, compared with $5.60 for Brazil at a similar stage. Good would be Mexico growing deposits, loans, and customer activity without needing promotional rates that burn the economics.

The recent online conversation is positive but thin. A 30-day scan found 20 useful items, led by discussion of the billion-dollar profit quarter, Mexico, and valuation. Treat the crowd as temperature, not evidence.

The numbers, translated

• Customers reached 138.9 million, up 13%, and the activity rate was 83.5%. Good is double-digit customer growth with activity above 80%, so Nu clears both lines.
• Monthly revenue per active customer reached $17.10, up 22%. This measures what Nu earns from each person who actually uses the platform. Good is growing faster than customer count, and 22% against 13% does exactly that.

• Gross revenue reached $5.88 billion, up 39%. This includes credit income, deposit-float income, and fees before direct costs. Good is growth above 25% without credit quality cracking, so the top line is exceptional.

• Net income reached $1.06 billion, up 49%. This is the official bottom line after all expenses and taxes. Annualizing one quarter gives roughly $4.24 billion, but that shortcut assumes current credit and currency conditions persist.
• Return on equity was 33%. Nu generated about 33 cents of annualized profit for each dollar of shareholder capital. Good banking economics usually exceed the cost of capital by a wide margin, and 33% is elite.
• The efficiency ratio was 19.5%. Operating expenses consumed about twenty cents of net revenue. Lower is better. Good for a scaled bank is below 40%, so Nu's branchless model is doing the heavy lifting.
• Risk-adjusted net interest margin reached 12.4%, up from 9.5% in the first quarter. This measures lending spread after accounting for credit losses. Good is a stable double-digit result. The jump is excellent, but part came from normal seasonal improvement in missed payments.
• Loans reached $39.4 billion, up 37%, while deposits reached $45.3 billion, up 18%. Good means faster earning-asset growth without straining funding or capital.
• Early-stage problem loans improved to 4.8%, but loans more than 90 days late rose to 6.9%. The first number says new stress eased. The second says older missed payments kept rolling into the serious bucket. Good would be the 90-day figure flattening below 7% as higher-risk lending matures.
• Shareholder equity was $13.25 billion, or about $2.74 per basic share. At $14.55, the stock trades near 5.3 times book value. Good banks earn a premium when returns stay high, but five times book leaves no room for ordinary credit mistakes.
• The market value is roughly $70.3 billion. Against annualized second-quarter profit, that is about 16.7 times earnings. This is not insane for a 39% revenue grower, but it assumes Nu can defend margins while expanding credit and new countries.

What I like

The cost advantage is real. Nu serves an active customer for roughly $1 per month without branches or legacy systems. Customers who use Nu as their primary bank provide richer data, and management says those customers show delinquency around half the portfolio average. That is the flywheel.

Mexico is moving from promise to infrastructure. The full-bank launch adds products, while a 35% loan-to-deposit ratio leaves room to grow lending.

Equity rose to $13.25 billion from $11.29 billion at year-end even after $500 million of repurchases. Profit now finances growth.

What worries me

Nu intentionally expanded into higher-risk lending. Early delinquencies improved, but late delinquencies climbed to 6.9%. If that keeps rising while loans grow 37%, seasonality stops being enough.

Paying 5.3 times book value assumes the 33% return on equity endures. If returns settle near 25% or credit costs rise, the multiple can compress while the company grows.

Brazil still carries the machine. Nu is diversified by country in customer count, but Brazil produced about 91% of the geographic revenue disclosed in its statutory note for core products. Mexico and Colombia need to become profit engines, not attractive slides.

Founder control is concentrated. Class B shares carry 20 votes each, against one for Class A. Public owners get little leverage when strategy goes wrong.

The United States is another claim on attention. Nu promises to keep initial spending below one efficiency-ratio point through 2027. Good means no larger bet before demand appears.

What would make this a buy

The price door opens around $12 to $13. At $12.34, the stock would trade near 4.5 times June book value. At $13.10, it would equal about 15 times annualized second-quarter earnings. That range still pays for quality, but it gives the credit cycle room to breathe.

The proof door opens if revenue grows above 30%, return on equity stays above 30%, efficiency remains below 22%, and 90-day problem loans hold below 7% for two quarters. Mexico's deposits and activity must also rise.

If the proof arrives first, I can pay more. If the price arrives because late delinquencies explode, I do nothing. Cheap and broken are not synonyms.

What makes me walk away

I walk if 90-day problem loans exceed 8%, or risk-adjusted margin falls below 9% while loan growth stays above 25%.

I walk if return on equity stays below 25% for two quarters while valuation remains above 5 times book.

I walk if Mexico's bank launch requires deposit rates that erase funding improvement, or if U.S. spending breaks management's promised efficiency budget before product-market fit appears.

Earnings On Deck

• MongoDB (MDB), September 1 after the close. Watch cloud-database consumption and whether AI workloads accelerate Atlas revenue. Good means revenue above roughly $735 million with guidance holding growth above 20%.
• UiPath (PATH), September 3 after the close. Watch revenue above $400 million, repeat-contract growth, and whether agentic automation moves from demos into paid production.
• Zscaler (ZS), September 3 after the close. Watch the first formal fiscal 2027 outlook. Good means revenue and annual recurring revenue growth land above the previously discussed 16% to 17% range.

IPO Watch

Gix Internet (GIXI) is expected to begin Nasdaq trading on September 1, but this is a dual listing of an Israeli public company, not a conventional U.S. IPO raising a large new pool of money. The accessible filing showed 2025 revenue of only about $7.7 million, and no traditional offer price range was disclosed. Good would be stable revenue and enough daily trading volume to create honest price discovery. This is a micro-cap listing, not a clean institutional debut, so stay away from opening-day noise.

The verdict

Score: 84 out of 100, candidate. Nu earned points for 139 million customers, 39% revenue growth, 49% profit growth, a 33% return on equity, a 19.5% efficiency ratio, rising customer monetization, cheap digital service, and a credible Mexican expansion runway. It lost points for a 6.9% late-stage problem-loan ratio, intentional movement into riskier credit, heavy Brazilian dependence, founder voting control, currency exposure, and a valuation near 5.3 times book value.

The bucket is candidate. The business clears the bar. The entry does not. Put it on the priority list and wait for $12 to $13, or pay more only after two quarters prove late-stage credit has stopped worsening.

The DBTech read

Risk-adjusted net interest margin is what a lender earns on loans after funding costs and expected credit pain. Nu's 12.4% says the lending engine kept a wide spread even after accounting for borrowers who may not pay. Think of a restaurant measuring profit after including spoiled food, not before. Good means the margin stays strong across seasons and riskier customer groups. A one-quarter jump is useful, but a stable year is proof.

Sources and as-of date

Research completed September 1, 2026 at 7:12 a.m. Eastern.

The $14.55 price came from Alpaca's IEX broker feed at 3:59:59 p.m. Eastern on August 31, 2026. IEX is narrower than the consolidated market, so small differences from other closing-price sources can occur.

Second-quarter operating metrics, management figures, Mexico's bank launch, product commentary, and the earnings release came from Nu's Form 6-K filed with the U.S. Securities and Exchange Commission on August 13, 2026: https://www.sec.gov/Archives/edgar/data/1691493/000129281426004222/nupr2q26_6k.htm

Statutory income, balance sheet, equity, share count, capital ratios, geographic disclosure, and subsequent events came from Nu's reviewed interim financial statements filed on August 13, 2026: https://www.sec.gov/Archives/edgar/data/1691493/000129281426004225/nufs2q26_6k.htm

Earnings dates came from UiPath and Zscaler investor-relations pages plus the Yahoo Finance calendar, retrieved September 1, 2026. Gix Internet's listing structure came from its Form 20-F filed with the SEC on June 18, 2026: https://www.sec.gov/Archives/edgar/data/1782265/000149315226029333/form20fr12b.htm

Community research covered Reddit, YouTube, GitHub, Hacker News, jobs, and prediction markets from August 2 through September 1, 2026. It returned 20 items across Reddit, YouTube, and GitHub, with no useful Hacker News, jobs, or prediction-market evidence. The sample was used for sentiment, not as proof of business quality.