One-to-Many Products, One-to-Many Countries: Scaling Nubank to 127 Million Customers

Cloud-native tooling and platform engineering promise everything we need to run software at scale: public clouds, infrastructure as code, developer tooling, and well-understood deployment and scaling models that abstract away complexity.

But what happens when business growth outpaces architectural assumptions?

This talk tells the story of how Nubank’s architecture evolved over a decade to support exponential growth, not just in traffic, but in products, regulatory domains and as an engineering organization.

When Nubank expanded from a single product and market into many products and many countries, the generic cloud deployment models and scaling patterns that served us at first became constraining.

We will walk through the challenges that emerged as:

• The number of products grew rapidly, from credit cards to bank accounts, loans, insurance, investments and more.
• Regulatory boundaries multiplied across countries.
• Our horizontal scaling strategy became the bottleneck.

Against this backdrop, we’ll explore some architectural adaptations:

• Internal architectural concepts to better reason about the system and enable flexible cloud deployment aligned with business needs.
• System boundaries to reduce blast radius and increase autonomy.
• Offering platform capabilities as products, hiding complexity from product engineers while providing opinionated “golden paths.”

You’ll leave with practical deployment patterns and concrete examples showing how architectural choices both enabled and constrained business outcomes. Expect to gain insight into treating platform architecture as a first-class strategic concern, not just an operational one, especially in organizations where growth introduces new dimensions of complexity.