Inter & Co added 7 million clients in 2025, while Nu Holdings pushed its nuFormer AI credit model into production across Brazil—then watched growth stall when regulators changed lending rules. The clash between algorithmic speed and regulatory reality is testing whether emerging market fintechs built durable businesses or optimized for conditions that no longer exist.
Nu's November 2025 stumble mirrors challenges facing AI lenders globally. When Brazil altered FGTS payroll loan regulations, the company's portfolio growth rate fell from 13-14% to roughly 10%. The AI model, trained on stable regulatory patterns, continued approving profiles that no longer performed under new constraints.
Inter projects its 2026 cost of risk will hit 5.5-6%, with private payroll delinquency exceeding 10%. Both banks report newer customer cohorts transacting faster than older groups—a pattern that either proves AI finds creditworthy borrowers traditional banks miss, or signals mispriced risk surfacing as growth compounds.
The divergence from developed market lending is structural. Western banks blend human judgment with algorithmic tools, diluting single-model failures. Brazil's digital-native lenders run pure AI underwriting across millions of daily micro-decisions, creating binary outcomes: either they capture opportunities competitors overlook, or they accumulate correlated exposures that deteriorate simultaneously under stress.
Nu's nuFormer enables feature rollout and customer scaling impossible with manual review—the company launched over 100 products in 2025. Inter's 7 million client gain proves market appetite for instant approvals. The question is whether speed-to-market traded away resilience.
Traditional underwriters adjust heuristics intuitively when macro conditions shift. Algorithms require retraining cycles that lag employment changes, government program modifications, and regulatory updates. Nu typically sees 15-90 day NPLs rise in Q1, indicating sensitivity to external shocks beyond model training data.
Emerging market fintech growth now depends on model adaptation, not just model accuracy. Banks that scaled portfolios during 2024-2025's favorable conditions will learn in 2026 whether their AI navigates headwinds—or if algorithmic underwriting created concentrated risk that human oversight would have caught.
Sources:
1 Globe Newswire, "Olympians Inspire Expands School Assembly and Leadership Workshop Programming Featuring Elite Athlet" (March 23, 2026)
2 Yahoo Finance, "Asian shares decline as hopes dim for resolution in Iran after Trump's latest comments" (March 23, 2026)
3 Yahoo Finance, "Varex Imaging Conference: CFO Highlights Tariff Pass-Through, India Factories, Cargo Systems & P" (March 22, 2026)
4 News Report, "Claro moves to acquire controlling stake in Desktop in $750M deal" (March 22, 2026)
5 Yahoo Finance, "NU's Efficiency Edge: A Key Factor Driving Its Premium Narrative" (March 20, 2026)