Twenty years of Amazon S3 and building what’s next

This article celebrates 20 years of Amazon S3, marking its launch on March 14, 2006, and discussing its evolution and future direction.

• S3 launched with simple PUT/GET primitives guided by five core fundamentals: security, durability, availability, performance, elasticity
• Scale grew from 1 petabyte to 500+ trillion objects across 123 Availability Zones in 39 regions
• Maximum object size increased 10,000x from 5GB to 50TB
• Price dropped 85% from 15 cents to 2 cents per gigabyte since 2006
• S3 API became industry standard; code written in 2006 still works unchanged today
• Engineering innovations include microservices for durability auditing, formal methods for correctness verification, Rust rewrites for performance
• Recent launches: S3 Tables for analytics, S3 Vectors for AI/semantic search, S3 Metadata for data discovery
• Vision: universal foundation for all data and AI workloads without moving data between systems

S3 has evolved from simple object storage to a foundational platform for modern data and AI workloads while maintaining backward compatibility and core principles.