Nu Holdings: The $28 Question Hiding in Plain Sight

Here is the data: 139 million customers. $1 billion in quarterly net income. That works out to roughly $7.20 per customer, per quarter. Annualized, that is $28.80 per head. For a digital bank in Latin America, those numbers look like a license to print money. But I have been staring at the unit economics of fintech for a decade, and I can tell you exactly where this story gets complicated. Let's be clear about what Nu Holdings (NYSE: NU) actually is. It is not a crypto company. It is not a blockchain protocol. It is a fully licensed Brazilian bank (Banco Múltiplo) that happens to run on modern cloud infrastructure. The crypto angle here is indirect but material: Brazil's central bank is rolling out DREX, a wholesale CBDC with smart contract capabilities, and Nu is one of the largest participants in the Pix instant payment rail that DREX will eventually settle on. If you are trading the intersection of traditional finance and digital assets, Nu is a bellwether for how Latin America's regulatory environment will treat programmable money. I have audited enough protocols to know that the real value in Nu is not the brand or the customer count. It is the stack. Nu is cloud-native, microservices-based, and operates with zero physical branches. That is not a cost advantage; it is a structural advantage. A legacy bank in São Paulo needs to amortize real estate, branch staff, and legacy mainframe maintenance across its customer base. Nu's marginal cost of serving a new customer approaches zero. That is why it can generate $1 billion in quarterly net income while charging fees that undercut Itaú and Bradesco by an order of magnitude. But here is the contrarian angle that most retail investors miss. The market treats Nu's 139 million customers as a monolithic growth asset. It is not. The critical metric is not customer count; it is wallet share. How many of those 139 million accounts are primary accounts where salaries are deposited? How many are dormant secondary wallets that users opened for a single Pix transfer and never touched again? Based on my experience analyzing user behavior in emerging markets, I would estimate that 30-40% of Nu's customer base is low-activity. That means the real active user base is closer to 90 million, and the effective ARPU on those active users is higher than the headline number suggests. The market is pricing Nu as a consumer bank, but it is actually a data company with a banking license. The data moat is real. Every transaction, every credit application, every behavioral signal feeds into Nu's machine learning models. This is the same flywheel I saw in the early days of on-chain credit scoring: more data leads to better risk pricing, which leads to lower defaults, which allows for more aggressive customer acquisition. Nu has been running this loop for years, and it shows in the credit quality. The NPL ratio remains manageable despite serving a customer base that traditional banks consider subprime. That is not luck. That is the compounding effect of proprietary data. Now, the risk side. I lived through the Terra collapse in 2022, and I learned that leverage cuts both ways. Nu's profitability is highly sensitive to Brazil's Selic rate. The current high-rate environment is a tailwind for net interest margins, but it is also a sword hanging over the credit book. If Brazil enters a recession, the C-class customers that Nu serves will be the first to default. The $1 billion quarterly net income figure is a point-in-time snapshot, not a guarantee. I have seen this movie before: high-growth lenders look invincible at the top of the cycle and look reckless at the bottom. The second risk is DREX itself. Nu is deeply integrated with Pix, and DREX will eventually replace or augment the settlement layer. If DREX enables smart contract-based payments, Nu's infrastructure is well-positioned to offer programmable financial products that traditional banks cannot match. But if the central bank imposes strict rules on DREX usage, or if it decides to compress interchange fees, Nu's cost structure could shift overnight. This is the same regulatory uncertainty that plagued the restaking protocols I audited in 2023: the technology is sound, but the rules are not yet written. The third risk is competitive. Mercado Pago, backed by the Mercado Livre e-commerce ecosystem, is Nu's most dangerous rival. Mercado Pago has something Nu lacks: an owned distribution channel with high-frequency engagement. E-commerce transactions happen daily; banking transactions happen weekly. That frequency advantage gives Mercado Pago a data edge that Nu cannot easily replicate. Nu is fighting back with wealth management and insurance products, but those are lower-frequency services. The battle for the Brazilian consumer's financial life is being fought on engagement, not product breadth. Let me give you a concrete signal to track. Nu's international expansion into Mexico and Colombia is the single biggest swing factor for the stock over the next 24 months. If Nu can replicate even 50% of its Brazilian success in Mexico, the valuation ceiling moves significantly higher. But if the international operations remain a drag on margins, the market will punish the stock. I have seen this pattern in every fintech expansion I have analyzed: the home market success creates overconfidence, and the international market exposes weaknesses in localization, regulatory navigation, and cultural fit. My takeaway is straightforward. Nu is a high-quality asset in a region with structural tailwinds. The unit economics are healthy, the data moat is real, and the management team has demonstrated execution discipline. But the current valuation assumes a smooth continuation of the Brazilian growth story, and that assumption is fragile. I would not chase the stock at current levels. I would wait for either a pullback driven by macro fears or a clear signal that the international expansion is gaining traction. The $28 per customer question is not about whether Nu can maintain that number; it is about whether it can grow it. Watch the ARPU trend, watch the Mexico customer count, and watch the DREX rulebook. Those three variables will determine whether Nu is a compounder or a value trap.