My takeaways from $NU Q4’25 earnings + call. TL;DR: The fundamental thesis remains intact. Today’s stock reaction is a head-scratcher. At ~17.5x 2026 EPS, with earnings likely (conservatively) compounding north of 30% over the next 3 years, the risk/reward here looks compelling. Let’s break it down. The core operating story remains strong. At a high level, key revenue drivers and profitability metrics were solid across the board: • Customers: 131mm (+22% CAGR since Q4’21) • ARPAC: $15 (+30% CAGR FXN over same period) • Deposits: +29% YoY FXN • Loan book: +40% YoY FXN • NPLs: healthy & stable • Efficiency ratio: at an all-time low • Cost of deposits (% of interbank): at an all-time low • ROE: 33% (ATH) Loan-to-deposit ratio sits at 50% → significant funding runway. Capital position: ~$9bn on $75bn of assets, ~$5.5bn above regulatory minimums. Hard to paint a much cleaner fundamental picture. So why did the stock sell off? Before the call, shares were up >5%. The tone shifted during the discussion around expenses. Three key nuances: 1. Efficiency ratio guidance: Management flagged 80–100bps upward pressure over the next 4–6 quarters. Drivers: • Return-to-office costs (largely pre-booked in Q4) • Strategic hiring/investment in US, Mexico, Colombia • Global expansion positioning Even at the high end, the efficiency ratio would be roughly back to Q3’25 levels — still best-in-class. These are growth investments with expected positive ROI, not structural cost creep. 2. Credit Loss Allowances (CLAs) + Risk-Adjusted NIM: CLA +26% FXN → Risk-adjusted NIM dipped to 10.5% from 10.8% QoQ. On the surface, that looks concerning. Mechanically, it isn’t. The driver: credit limits expanded from ~$18bn to ~$29bn. Under accounting rules, $NU must provision on the limits, not just funded loans. So provisioning rose largely because of higher approved capacity — which itself reflects improved underwriting (AI-driven programs implemented over the past 18 months). Additionally: • Mexico net income absorbed a one-off $25mm deposit insurance contribution. • Ex that, risk-adjusted NIM would’ve been broadly stable QoQ. 3. Q4 accounting noise: • +$58mm net income from deferred tax asset remeasurement (Brazil rate changes). • -$25mm deposit insurance impact (Mexico). • -$22mm pre-provisioned RTO expenses. Net effect at the EBT level: largely a wash. Headline numbers look messy. Underlying economics did not change. Markets often extrapolate misunderstood one-offs into perpetuity. In a stock with heavy retail + momentum participation, one large bearish flow can trigger reflexive machine-driven selling — especially when surface-level metrics look “worse.” But zoom out. $NU unit economics remain exceptional: • Cohort ARPAC >$30 at 96 months • Stable cost to serve • Underutilized balance sheet • Embedded operating leverage • Strong underwriting discipline • Optionality from new geographies The long-term compounding engine appears intact. At this valuation, with this growth profile, I continue to view $NU as a compelling long-term buy-and-hold.