Sunday, November 22, 2026

Brazilian Digital Banks Hit 65.6% CDI Funding Costs, 20 Points Below Traditional Lenders as AI Credit Models Drive Global Fintech Gap

Inter & Co recorded 65.6% CDI funding costs in Q4 2024, 15-20 percentage points below Brazil's traditional banks, highlighting the global divide between AI-driven and conventional lenders. Nu Holdings and Inter deployed production machine learning credit models in 2025, processing applications through systems trained on over 100 million data points. The efficiency gap mirrors trends in India, Southeast Asia, and Latin America where digital banks use algorithmic underwriting to underprice establi

From our archive, published before source-traceability. How we work

Brazilian Digital Banks Hit 65.6% CDI Funding Costs, 20 Points Below Traditional Lenders as AI Credit Models Drive Global Fintech Gap
Image generated by AI for illustrative purposes. Not actual footage or photography from the reported events.

Inter & Co recorded 65.6% CDI (Certificado de Depósito Interbancário) funding costs in Q4 2024, the lowest rate among Brazilian financial institutions serving over 20 million clients. Traditional banks in Brazil operate 15-20 percentage points higher, a gap comparable to digital-versus-traditional spreads in India's UPI-based lenders and Southeast Asian neobanks.

Both Inter and Nu Holdings deployed production AI credit models during 2025. Nu's nuFormer system processes loan applications using machine learning trained on over 100 million customer data points. The technology maintains risk-adjusted net interest margins despite government FGTS (employment guarantee fund) program changes that reduced industry margins, according to CFO Guilherme Lago.

Inter's newer client cohorts transact faster and more frequently than earlier users, indicating improved targeting from algorithmic underwriting. The bank became Brazil's fastest-growing financial institution among those exceeding 20 million clients, growth correlating with real-time risk pricing unavailable to competitors without comparable AI systems.

The funding cost advantage creates a compounding cycle: lower deposit rates improve margins, funding customer acquisition, which generates more training data for credit models. This flywheel allows digital banks to underprice traditional lenders while maintaining profitability, a pattern visible across emerging markets from Jakarta to Lagos.

Brazil's Central Bank requires identical minimum capital ratios regardless of underwriting technology. AI models help digital banks optimize within these constraints by identifying lower-risk borrowers who receive identical treatment under traditional methods. The regulatory framework mirrors approaches in Mexico, Colombia, and other Latin American markets balancing fintech innovation with systemic stability.

The 2026-2027 period will test whether AI credit advantages survive economic cycles. Brazilian interest rates remain volatile, and machine learning models trained during growth periods may underperform in downturns. Non-performing loan rates will determine if current efficiency gaps reflect permanent structural advantages or temporary market conditions, with implications for digital banking strategies across developing economies where similar AI deployment is accelerating.


Sources:
1 Yahoo Finance, "Why The Roche Holding (SWX:ROG) Narrative Is Shifting On Mixed Pipeline Signals And Valuation" (March 22, 2026)
2 Yahoo Finance, "How Vertex’s IgA Nephropathy Data and Accelerated FDA Pathway At Vertex Pharmaceuticals (VRTX) Has C" (March 22, 2026)
3 Yahoo Finance, "NU Q4 Earnings Call Highlights" (February 28, 2026)
4 Yahoo Finance, "Nu Holdings Ltd (NU) Q4 2025 Earnings Call Highlights: Record Growth and Strategic Expansion" (February 28, 2026)
5 Yahoo Finance, "IP Group PLC (IPZYF) Full Year 2025 Earnings Call Highlights: Strong NAV Growth Amid Market ..." (March 23, 2026)

Are you or your business in this article?

If something is wrong, out of context or missing, tell us. We correct errors with a dated note, and people named in an article can publish their response under it.

Correct, add context or respond →

Not a source? Spotted an error? Report an error

How this story was made

Spotted an error? Use “Correct this article” (bottom right), or select any sentence, to report it or respond to that point.

  1. Published before we listed source documents with each article.
  2. Published 7 Mar 2026
  3. No responses yet from the people or organisations in this story. Are you in this story? Respond on the record ›
  4. Part of the Via News record, fingerprinted daily so it cannot be quietly rewritten. What that means › · Report an error ›
ViaNews Editorial Team

Articles published by the Via News desk since 2015: news, explainers and video features, under the editorial responsibility of L. Miguel Salvado, Editor. Via News is published by VIA GLOBE UNIPESSOAL LDA, Lisbon, Portugal.