Will Nu Holdings' consolidated 90+ NPL ratio exceed 8.0% by Q2 2026?

Current NPL at 6.6% would need a 140bps increase in two quarters to breach 8.0%. While the $29B unused credit limits represent latent exposure and AI models are untested through EM downturns, Brazil's economy is not currently in recession and the 2x funding coverage provides significant absorption capacity. Historical EM credit cycles show NPL deterioration is typically gradual unless triggered by a macro shock. No such shock is currently evident.

The shift from Brazil-only to consolidated NPL reporting makes the 6.6% baseline less reliable — it could be masking worse performance in newer markets (Mexico, Colombia). The $29B in unused limits (60% YoY growth) driven by AI models creates non-linear risk if utilization spikes. However, the question asks about Q2 2026 — only 2 quarters away — and credit deterioration of this magnitude typically requires a macro catalyst that is not yet visible. Brazilian Selic at ~14% supports credit performance through high deposit returns.

Three key factors argue against a rapid NPL spike: (1) management's credit-first strategy with conservative provisioning under IFRS 9, (2) the AI credit decisioning has been gaining market share at the fastest rate in 10+ quarters suggesting the models are working, and (3) $3.6B in regulatory capital plus $2.2B excess provides deep buffers. The 140bps jump would imply significant portfolio stress that would likely be preceded by leading indicators (delinquency trends, utilization spikes) that management would respond to by tightening credit standards.

The 8% threshold is aggressive for a 2-quarter timeframe. Current NPL trend is stable-to-improving. Brazil's economy faces headwinds from high Selic but this actually helps credit quality by keeping deposit/float income high. The unused limits are a medium-term risk, not a Q2 2026 risk absent a macro shock. Management's track record of credit quality management has been strong.

The consolidated NPL switch is concerning — without a bridge to prior Brazil-only data, we cannot assess the true trajectory. Mexico and Colombia portfolios could be materially weaker. The 60% expansion in unused limits is aggressive by any standard. However, even accounting for these unknowns, a 140bps deterioration in 6 months requires a specific credit event. Assigning slightly higher probability due to data opacity.

Base case is NPL stays in the 6-7% range through Q2 2026. Brazil is growing at ~2-3% GDP, no recession in sight. EM credit cycles can turn quickly but typically need a trigger: currency crisis, commodity shock, or political instability. None of these are imminent for Brazil. The 2x funding coverage and $3B unrestricted cash mean Nu has tools to manage credit stress even if it emerges.

6.6% to 8.0% in two quarters without a recession is historically uncommon for well-managed credit portfolios. Nu's AI models and 2x funding coverage reduce risk further. Low probability.

Some risk from NPL opacity and AI model untested nature, but timeline is too short for this magnitude of deterioration without a macro shock. Unused limits create medium-term tail risk but not a Q2 2026 probability of 8%+ NPL.

Current 6.6% with stable trajectory, 2x funding coverage, and IFRS 9 provisioning. No visible catalyst for rapid deterioration. Credit-first strategy with AI decisioning has been working. Low probability of 8%+ in Q2.